"Features" are values derived from the user’s data, which can be used directly in a loan decisioning flow. They are returned in the decision endpoint. Definitions of the features generated by Pngme are:
| Feature | Definition | Use Case | Value Proposition | Response value |
|---|---|---|---|---|
| average_end_of_day _depository_balance _{t0_days}_{t1_days} | Average of daily total balance held in all depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Increase customer retention and loyalty by offering tailored solutions based on their balance needs and preferences • Enhance customer acquisition and cross-selling by identifying potential customers with high balance levels and offering them attractive incentives and rewards • Improve revenue and profitability by optimizing the interest rates and fees based on the balance segments and risk profiles • Reduce operational costs and risks by streamlining the processes and systems for managing the balance data and ensuring its accuracy and security | • Help banks and fintechs to understand the financial behavior and liquidity of their customers • Help financial institutions to segment their customers based on their balance levels, offer personalized products and services, and optimize their pricing and risk strategies. | {float, null} |
| average_end_of_day _loan_balance _{t0_days}_{t1_days} | The time-average of the user’s balances, for all balances observed in SMS from known lenders, across all known lenders. The average is calculated over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Reduce credit risk and losses by identifying customers who are over-indebted or have high default probability across different lenders • Increase customer satisfaction and loyalty by offering debt consolidation or refinancing options to customers who have multiple loans with high interest rates or fees • Enhance customer segmentation and targeting by understanding the debt profile and behavior of customers across different lenders and offering them relevant products and services • Improve data quality and reliability by using SMS data as a source of truth for verifying the balances and transactions of customers across different lenders | • Measure the debt burden and repayment capacity of their customers • Monitor the credit performance and delinquency of their customers across different lenders. | {float, null} |
| count_airtime_purchase_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user purchased airtime with any telco. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Increase customer engagement and retention by offering mobile-related products and services, such as mobile banking, mobile money, or mobile insurance • Enhance customer segmentation and targeting by understanding the mobile preferences and behavior of customers, such as their preferred telco, airtime amount, and frequency of purchase • Improve data quality and reliability by using SMS data as a source of truth for verifying the mobile transactions and activities of customers | • Understand the mobile usage and spending patterns of their customers • Gauge the customer satisfaction and loyalty with their current telco provider. | {int, null} |
| count_betting_and_lottery_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in betting or lottery activity. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance the ability to manage risk by providing additional data points for assessing customer's financial stability • Offer personalized services such as financial planning advice or tools to customers who frequently engage in betting or lottery activities to help them manage their finances better • Improve customer service and build stronger relationships with customers, thereby increasing customer retention by understanding customers’ habits and preferences | • Assess the risk profile of a potential borrower as a user who frequently engages in betting or lottery activities might be considered a risk • Incorporate this data into credit scoring models to help determine a user's creditworthiness • Understanding a user’s betting or lottery activity can help tailor their marketing strategies and develop products that cater to this specific customer segment | {int, null} |
| count_[fastfood_events _categories]_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through food events, where food events are broken down into general kfc_, java, creamy inn, pizza inn, chicken inn,galitos events_. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer service and build stronger relationships with customers, thereby increasing customer retention by understanding customers’ spending habits and preferences on food events. • Use data to offer personalized services such as budgeting and financial planning tools that track food events and compare pricing and packages from different catering or food precessing services, helping them make cost-conscious decisions and manage their finances better. • Improve data quality and reliability by using SMS data as a source of truth for verifying food events of customers. | • These features may be indicative of individuals who frequently dine out, possibly suggesting a personal or professional interest in the food and drink industry. • Understanding a user’s activity related to food events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_[financial_events _categories_]_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through financial events, where financial events is broken down into general financial spending, insurance, investment, SACCO and service events. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance customer engagement and retention by offering tailored product recommendations for financial events such as insurance, investment and savings services, helping them make cost-conscious decisions and manage their finances better. • Improve data quality and reliability by using SMS data as a source of truth for verifying financial events of customers. | • These features may be indicative of an individual’s interest in financial growth through investments and corporate savings. It may also point to individuals owning high-valued properties that necessitate insurance. • Understanding a user’s activity related to financial events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_[food_events _categories]_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through food events, where food events are broken down into general food spending, bar and restaurant, catering, food processor and distributor, food store, wine and spirit events. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer service and build stronger relationships with customers, thereby increasing customer retention by understanding customers’ spending habits and preferences on food events. • Use data to offer personalized services such as budgeting and financial planning tools that track food events and compare pricing and packages from different catering or food precessing services, helping them make cost-conscious decisions and manage their finances better. • Improve data quality and reliability by using SMS data as a source of truth for verifying food events of customers. | • These features may be indicative of individuals who frequently dine out, possibly suggesting a personal or professional interest in the food and drink industry. • Understanding a user’s activity related to food events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_golf clubs_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user uses golf clubs. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days | • Improve customer service and build stronger relationships with customers, thereby increasing customer retention by understanding customers’ spending habits and preferences luxury events • Use data to offer personalized services that are related to golf club events to improve customer satisfaction. • Improve data quality and reliability by using SMS data as a source of truth for verifying events of customers. | • These features may be indicative of individuals who frequently visit golf clubs and are interested in this industry. • Understanding a user’s activity related to golf events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_[health_events _categories]_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through health events, where health events are broken down into general health, hospital, pharmaceutical and specialist events. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance customer engagement and retention by offering tailored recommendations on healthy living, tips and HMOs. • Improve data quality and reliability by using SMS data as a source of truth for verifying health events of customers. | • These features may be indicative of the amount spent on health-related services, potentially highlighting individuals facing health challenges who could be targeted for health product campaigns. • Understanding a user’s activity related to health events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_insufficient_funds_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user had insufficient funds to conduct a transaction. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve decision making by integrating real-time insights to the decision-making process, particularly in credit risk assessment and loan approval • Improve customer satisfaction and loyalty by offering overdraft protection services or low-balance alerts to customers who frequently have insufficient funds • Enhance customer experience by proactively reaching out to customers who frequently receive insufficient funds notifications and offer solutions or programs to help them better manage their funds • Reduce the risk of loan defaults and overdrafts by setting lower credit limits, requiring collateral for loans, or offering financial counseling | • Assess the credit risk of a customer based on their frequency of insufficient funds notifications • Segment customers based on their financial behavior for more targeted marketing and product development | {int, null} |
| count_loan_declined_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user was declined for a loan, from any known lender. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Accurately assess the creditworthiness of customers who have been declined for loans in the past • Target users who have been declined for loans in the past with marketing campaigns that are tailored to their needs • Develop new products and services that are specifically designed for users who have been declined for loans in the past • Protect customers from fraudulent loan applications • Mitigate the risk associated with lending to users who have been declined for loans in the past | • Understand the borrowing behavior of a customer • Use this data to segment their market and tailor their products and services accordingly • Gain insights into the lending criteria used by other lenders in the market, helping banks and fintechs to stay competitive | {int, null} |
| count_loan_defaulted_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user defaulted on a loan, from any known lender. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance credit scoring algorithms and use it to adjust credit scores based on the user's history of loan rejections and provide more accurate risk predictions • Use the data to offer personalized financial counseling services and reach out to individuals who have experienced loan rejections, providing guidance on improving financial health, managing debt, and increasing creditworthiness • Use the data to identify gaps in the market and develop innovative financial products tailored to the needs of individuals who have been declined for loans | • Enhance risk assessment model by understanding the frequency of loan rejections from other lenders • Perform competitor analysis and identify trends in rejection rates, assess the competitive landscape, and make informed decisions on loan approval criteria • Gain users insights into the reasons behind loan rejections | int |
| count_loan_missed _payment_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user missed a loan payment, for a loan from any known lender. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Use the data to build predictive models that forecast the likelihood of a user defaulting on a loan • Customize loan terms based on the user's history of missed payments such as offering flexible repayment schedules or alternative payment plans to accommodate the financial situation of individuals • Optimize collection strategies by segmenting users based on their missed payment history • Offer credit counseling services to users who have a history of missed payments | • Gain insights into customers financial behavior and make more informed decisions on loan approvals • Optimize their collection strategies by leveraging information on missed loan payments • Integrating information on missed loan payments into credit scoring models enhances their accuracy | {int, null} |
| count_loan_opened_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user either had a loan approved and/or had a loan disbursed, from any known lender. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Offer relevant products or services to users who are likely to have specific financial needs following a successful loan transaction • Tailor product offerings based on the type of loans recently approved or disbursed • Assess the user's capacity for additional credit based on their successful loan transactions and adjust credit limits accordingly • Evaluate market expansion opportunities by analyzing where successful loan approvals and disbursements are concentrated. | • Assess an individual's creditworthiness based on their history of successful loan approvals and disbursements • Identify users who have recently had a loan approved or disbursed and leverage this information for targeted cross-selling • Gain insights into popular loan types, preferred terms, and customer preferences, helping in strategic planning and product development. • Engage with customers who recently had a loan approved or disbursed, offering support, financial advice, or exclusive benefits to strengthen the customer-bank relationship | {int, null} |
| count_loan_repayment_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user made a loan repayment to one of their loans, from any known lender. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Assess a user's capacity for additional credit and offer credit limit adjustments accordingly • Provide personalized loan terms, interest rates, and product features to users who have demonstrated responsible repayment behavior • Launch financial education initiatives targeted at users who have made consistent loan repayments • Use data on loan repayments to enhance default prediction models | • Identify low-risk customers and adjusting risk assessment models for more accurate lending decisions • Recognize and reward users who consistently make timely loan repayments • Use data on loan repayments to adjust credit scores dynamically | {int, null} |
| count_mpesa_buygoods_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through M-Pesa BuyGoods. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer satisfaction and loyalty by offering customers access to exclusive shopping platform events, such as flash sales, early access events, and member-only discounts, rewarding their loyalty and encouraging repeat purchases. • Improve data quality and reliability by using SMS data as a source of truth for verifying buygoods events of customers. | • This feature may be indicative of an individual’s general spending power on goods and products, reflecting their economic capability to purchase various items. • Understanding a user’s activity related to M-Pesa BuyGoods can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_mpesa_p2p_events _{t0_days}_{t1_days} | Total number of M-Pesa transactions that were conducted from peer to peer, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Offer promotions on certain products or services, providing added value to users with higher M-Pesa transactions conducted from peer to peer. • Provide insights and recommendations to help users optimize their financial strategies, such as suggesting investment options or debt management tips | • Segment customers and offer targeted financial products and services where users with higher M-Pesa transactions conducted from peer to peer may be eligible for premium services or personalized offerings • Refine credit scoring models, allowing for a more accurate evaluation of a user's financial capacity • Understanding a user’s activity across M-Pesa transactions conducted from peer to peer can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_mpesa_paybill_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through M-Pesa Paybill. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer satisfaction by recommending platforms that provide ease of payment for bills • Improve data quality and reliability by using SMS data as a source of truth for verifying paybill events of customers. | • This feature may be indicative of an individual’s general spending power on bills and services, reflecting their economic capability to service various bills. • Understanding a user’s activity related to M-Pesa Paybill can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_of_currencies _{t0_days}_{t1_days} | Total count of currencies the user transacted with over the previous t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Automate currency conversion processes based on the identifieds currencies, providing users with real-time and accurate exchange rates • Tailor product recommendations, such as savings accounts or investment options, based on the user's identified currencies and financial goals • Facilitate international transactions by recognizing currencies, allowing for seamless cross-border payments and reducing currency-related complexities | • Improve transaction processing efficiency by automatically identifying the currencies, reducing errors and providing a seamless financial experience • Facilitate cross-border financial planning by recognizing the user's currencies, enabling more accurate budgeting and investment decisions | {int, null} |
| count_of_institutions _{t0_days}_{t1_days} | The number of distinct financial institutions represented in the SMSs received by the user. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Provide personalized product recommendations based on the range of financial institutions users interact with • Tailor communication and engagement strategies to strengthen relationships with users who engage with a variety of financial institutions • Identify market expansion opportunities by analyzing regions or demographics where users interact with a diverse set of financial institutions. • Refine risk assessments, considering the diversity of financial relationships as a factor in determining creditworthiness. | • Identify users who engage with multiple financial institutions, presenting an opportunity for cross-selling • Evaluate the risk diversification of users by analyzing their interactions with different financial institutions | {int, null} |
| count_overdraft_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user had overdraft activity on an account. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Assess the creditworthiness of potential borrowers by using this data feature as part of their credit scoring models • Develop loan products that are tailored to the needs of customers who have a history of overdraft activity • Target customers who have a history of overdraft activity and offer them better loan products • Manage the risk of lending to individuals who have a history of overdraft activity | • Avoid bad debts and reduce the risk of default by mitigating the risk of lending to individuals who have a history of overdraft activity • Improve customer experience and increase customer loyalty by offering personalized loan products to customers who have a history of overdraft activity | {int, null} |
| count_[public_institutions _events_categories] _events_{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through public institutions, where public institutions are broken down into general public institutions, charity, church, government and school events. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | •Increase customer engagement and retention by personalized recommendations for public institution such as charity and churches based on their past donations and interests • Improve customer satisfaction and loyalty by offering loans for school fees pyment at the needed time • Improve data quality and reliability by using SMS data as a source of truth for verifying public institution events of customers. | • These features may be indicative of individuals belonging to a religious community and inclined towards charitable giving. It may also suggest individuals with dependents in educational institutions or those pursuing education themselves. • Understanding a user’s activity related to public institution events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_salary_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user received income through salary. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Use the data to build predictive models that forecast the likelihood of a user defaulting on a loan based on estimated monthly income • Improve data quality and reliability by using SMS data as a source of truth for verifying salary income of customers. | • These features may be indicative of the inflow an individual receives on a monthly basis, serving as useful data for estimating income. • Understanding a user’s activity related to Salary income can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_[shopping_events _categories]_events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through shopping events, where shopping events are broken down into general shopping spending, clothing and fashion, ecommerce platform, entertainment and subscription, professional service, retail store, supermarket, wholesale and distribution , naivas, quickmart, carrefour. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer satisfaction and loyalty by offering customers access to exclusive shopping platform events, such as flash sales, early access events, and member-only discounts, rewarding their loyalty and encouraging repeat purchases. • Enhance customer engagement and retention by offering tailored product recommendations, enhancing their shopping experience and increasing the likelihood of purchase • Improve data quality and reliability by using SMS data as a source of truth for verifying ecommerce activities of customers. | • These features may be indicative of an individual’s spending capacity across a broad spectrum of goods and services, offering insights into their overall financial behaviour and preferences. • Understanding a user’s activity related to shopping events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_transactions_depository _{t0_days}_{t1_days} | The number of SMS received indicating that the user conducted a depository transaction. The count is summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | optimising• Create customized savings plans for users based on their depository transaction history. • Introduce investment advisory services for users with significant depository transactions • Launch targeted marketing campaigns for specific user segments based on depository behavior | • Understanding how users engage with depository services helps in tailoring offerings and optimizing service delivery • Offer personalized financial planning services based on users' depository transactions • Gain insights into saving patterns, recommend investment opportunities, and assist users in achieving their financial goals. | {int, null} |
| count_[transportation_and_ travel_events_categories] _events_{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through transportation and travel events, where transportation and travel events are broken down into general transportation and travel spending, airfare, hotel, logistics, parking and petrol station events. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Increase customer loyalty and retention by offering incentives on transportation and travel events services, such as discounts, or redeemable purchase points on airfare, hotel etc. • Offer personalized services such as airfare events comparison across different airlines & dates or offer lodging options to customers who frequently engage in travel for smoother commute. • Improve data quality and reliability by using SMS data as a source of truth for verifying transportation and travel events of customers | • These features may be indicative of individuals who travel, commute, or vacation regularly, as well as those owning vehicles, suggesting a certain lifestyle or mobility requirement. • Understanding a user’s activity related to transportation and travel events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_user_shared _device_ids_0_10 | The number of distinct users using the same device updated within the time range of 10 days history prior to the prediction date. | • Implement multi-user authentication mechanisms for devices with multiple distinct users • Offer features that allow users to link their accounts on a shared device for a more integrated and collaborative financial management experience • Send alerts to users about multiple logins or transactions from a single device, providing an added layer of security and transparency • Analyze user behavior on shared devices to provide targeted offers and promotions that align with the collective financial needs and preferences of the users on that device | • Identify potential security risks and ensuring that only authorized users access sensitive financial services • Identify unusual patterns of multiple users on a single device, helping to flag potential fraudulent activities and protect users from unauthorized access • Tailor user experiences based on the number of distinct users sharing a device | {int, null} |
| count_user_shared _phone_numbers_0_10 | The number of distinct users with the same phone number updated within the time range of 10 days history prior to the prediction date. | • Implement multi-user authentication mechanisms for shared phone numbers • Offer features that allow individuals with the same phone number to link their accounts, facilitating collaborative financial management and coordination • Notify users about suspicious activities or transactions, enhancing security and ensuring that all relevant parties are informed | • Enhance identity verification and fraud prevention measures • Optimize communication channels, ensuring that messages and alerts are appropriately directed to all relevant users associated with a specific phone number | {int, null} |
| count_[utilities_events _categories] _events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through utilities, where utilities are broken down into general utilities spending, airtime, energy, housing, _kplc, **_internet, and water** events. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance customer engagement and retention by enabling customers to track their spend and usage patterns across different utility events inorder to manage consumption. • Improve customer satisfaction by recommending platforms that provide ease of payment for utility bills • Improve data quality and reliability by using SMS data as a source of truth for verifying utility events of customers. | These features may be indicative of individuals who earn regularly and possess sufficient resources to cover short-term bills monthly or weekly. Such individuals are likely to be relatively stable financially, as evidenced by their consistent bill payments. • Understanding a user’s activity related to utilities events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {int, null} |
| count_[vendors_events _categories] _events _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in pushing from suppliersgallantchoices spending, japasupplies, startbase enterprises, merryland stores, sweetbargain traders, kiodistributors, kopokopo, paystack, pesapal, sokowatch .The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | Identify the vendors used by small business owners • Get insights on the type of business one is engaging in. ease of payment for utility bills • Improve decision making when loaning to small business owners | these features indicate users who are likely to small business owners | {int, null} |
| daily_average_of_stacked _loan_alerts_0_90 | The daily average number of loan-related SMS from known lenders received by a user. Loan-related SMS include, LoanDefaulted, LoanMissedPayment, LoanRepaid, LoanApproved, LoanDisbursed, LoanRepayment, and LoanRepaymentReminder, over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Provide personalized loan recommendations based on the user's daily average engagement with loan-related SMS • Optimize the timing and frequency of loan-related communications to maximize effectiveness • Assess the potential risk associated with users who frequently interact with loan-related content and adjust risk management strategies accordingly | • Use data on the daily average number of loan-related SMS to assess a user's creditworthiness • Enable targeted marketing campaigns based on the user's engagement with loan-related SMS • Adjust credit scores to reflect the user's interaction patterns and level of interest in loan products • Analyze user engagement with loan-related content to understand preferences and trends | {float, null} |
| daily_average_of_stacked _loan_fes_0_90 | The daily average number of lenders with whom a user has activity over a period of zero to 90 days history prior to the prediction date. Having activity means receiving one or more SMS on a given day, from a given lender, showing an active loan (the SMS stating the outstanding loan balance and/or a disbursement/repayment event). The daily average number of lenders means the mean average of lenders on a per day basis, for all days in which activity was observed with one or more lender, over the last 0 to 90 days history prior to the prediction date. | • Adjust credit limits and assess a user's overall creditworthiness based on their engagement with various lenders • Create bundled offerings that provide users with comprehensive financial solutions, combining various products to meet their diverse needs • Explore new markets or demographics that show a higher propensity for engaging with multiple lenders • Offer incentives, rewards, or exclusive benefits to enhance user loyalty and satisfaction | • Use the daily average number of lenders with user activity to assess creditworthiness and evaluate risk • Enable targeted product recommendations based on the daily average number of lenders • Gain insights into market trends by analyzing user engagement with a variety of lenders | {float, null} |
| data_density_{t0_days}_{t1_days} | Ratio of active user days to total number of days in range, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Customize the user experience, displaying relevant information or features more prominently for users with higher engagement ratios • Offer incentives, promotions, or rewards to re-engage users who have been less active, encouraging them to explore additional services • Proactively address the needs of users whose engagement has decreased, offering personalized solutions to retain their loyalty • Implement usage-based fee structures for certain services based on user engagement ratios | • Understand how frequently users interact with financial services and products, enabling institutions to tailor their offerings accordingly • Categorize users into segments such as high-engagement, moderate-engagement, and low-engagement, allowing for personalized communication and product recommendations • Optimize the timing, frequency, and channels of communication, ensuring that messages are well-received by actively engaged users • Prioritize product development and features based on user engagement ratios | {float, null} |
| data_oldest_minutes | The time in minutes between utc_endtime and the oldest financial event or alert, as an indicator of data age, or earliest activity within a 180 day period. | • Actively track transactions which occured farthest to utc_endtime to identify potential anomalies or suspicious activities • Send relevant communications to users based on their earliest financial events such as trigger notifications, alerts, or reminders to keep users informed about loyalty offers. • Adjust risk scores based on the age of financial events, ensuring that risk assessments are reflective of the earliest user behavior • Target users financial activity legth, tailoring marketing messages and offers to align with their current financial needs and interests | • Leverage the time difference to dynamically update reporting and dashboards by working with the earliest financial data, providing a more accurate and timely picture of the current financial landscape • Set thresholds based on the age of financial events, flagging or investigating transactions that deviate from expected recency patterns • Send targeted communications, alerts, or offers to users, taking into account their earliest financial activities | {int, null} |
| data_recency_minutes | The time in minutes between utc_endtime and the most recent financial event or alert, as an indicator of data recency, or freshness. | • Actively track transactions occurring close to utc_endtime to identify potential anomalies or suspicious activities • Send timely and relevant communications to users based on their recent financial events such as trigger notifications, alerts, or reminders to keep users informed about their financial activities • Adjust risk scores based on the recency of financial events, ensuring that risk assessments are reflective of the latest user behavior • Target users with recent financial activities, tailoring marketing messages and offers to align with their current financial needs and interests | • Use the time difference as an indicator of data recency to make real-time decisions • Leverage the time difference to dynamically update reporting and dashboards by working with the latest financial data, providing a more accurate and timely picture of the current financial landscape • Set thresholds based on the recency of financial events, flagging or investigating transactions that deviate from expected recency patterns • Send targeted communications, alerts, or offers to users, taking into account their recent financial activities | {int, null} |
| difference_count_of_loans_opened _to_loans_delinquent _{t0_days}_{t1_days} | The difference between a) the number of SMS received indicating that the user was approved for a loan, or had a loan disbursed (of any loan to any lender) and b) the number of SMS received indicating that the user is past due or defaulted on a loan (of any loan to any lender). The counts for a) and b) are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Adjust approval criteria or streamline processes for users with a positive SMS difference, potentially offering faster approvals or better terms • Implement early intervention strategies, reaching out to users with negative differences to offer assistance, alternative repayment plans, or financial education • Assess a user's capacity for additional credit and offer credit limit adjustments accordingly • Identify patterns and make informed decisions about adjusting lending strategies | • Use the difference in SMS counts as an indicator of a user's creditworthiness - a positive difference may suggest responsible financial behavior, while a negative difference may signal potential credit risk • Adjust credit scores based on the net behavior, helping in identifying users with a strong credit history or those facing financial challenges • Provide personalized advice and support to users experiencing challenges, aiming to prevent future defaults and improve overall financial well-being • Tailor loan offerings, interest rates, and repayment terms to better suit the financial profiles of users with varying levels of credit risk | {int, null} |
| difference_count_of_loans_opened _to_loans_repaid _{t0_days}_{t1_days} | Returns the count of loans approved or disbursed - loans repaid, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. The number of SMS received indicating that the user had a loan approved or disbursed, minus the number indicating they repaid a loan. The counts are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Adjust their approval processes for users with different loan approval and repayment patterns, ensuring that lending decisions align with historical performance • Identify users who may benefit from repayment assistance programs • Refine credit scores to reflect user behavior in terms of loan approvals, disbursements, and repayments, enhancing the accuracy of credit assessments. | • Use the count of approved or disbursed loans minus repaid loans to analyze their lending portfolio's performance over different time windows • Incorporate the historical loan data into risk management models to forecast future repayment trends • Inform strategic product development by understanding the demand for loans over different time periods • Customize marketing messages and offers for users with specific lending behaviors, improving the effectiveness of marketing campaigns | {int, null} |
| is_mpesa_business | Identifies user as business or not, this decision is based on mpesa sms patterns that indicate a business eg. sms indicating money received by the MPESA buygoods service,MPESA pochi la biashara etc returns True if found. Also checks number of credit transactions in a month and mpesa credit fraction and based on the threshold returns True if criteria is met, returns False if none is of the conditions are met | •Offer tailored products and services such as business loans and online payment gateways to users identified as business. •Improved Risk Management accurate classification will help in setting appropriate credit limits and managing risks. •Leverage the classification to perform in-depth financial analysis specific to businesses, such as cash flow projections, financial health monitoring. | •By offering tailored products and services, the bank can cross-sell and up-sell more effectively. •The feature provides valuable insights into the financial behavior of business customers, enabling better decision-making and strategic planning. | {boolean} |
| loan_sharks_alert _institutions_ratio_0_90 | Ratio of loan shark related financial events over total events over a given period, over the last 0 to 90 days history prior to the prediction date. | • Implement real-time monitoring systems to detect unusual spikes in the ratio of loan shark-related events by investigating and taking prompt action to protect users and maintain the integrity of their financial services. • Provide information, warnings, and resources to help users understand and avoid engaging with loan sharks • Share relevant data and collaborate with authorities to investigate and combat illegal lending practices | • Use the ratio to identify the presence of loan shark institutions and mitigate associated risks • Ensure adherence to regulatory requirements, report any unusual patterns, and collaborate with regulatory authorities to address potential issues • Proactively inform users about the risks associated with loan shark institutions, providing guidance on how to recognize and avoid such entities • Incorporate the ratio into risk assessment models to refine credit scoring and lending decisions by adjusting risk parameters based on the prevalence of loan shark-related events, improving the accuracy of risk assessments | {float, null} |
| loan_sharks_fes_ratio_0_90 | Ratio of loan shark institutions over total seen institutions over the last 0 to 90 days history prior to the prediction date. | • Implement real-time monitoring systems to detect unusual spikes in the ratio of loan shark-related events by investigating and taking prompt action to protect users and maintain the integrity of their financial services. • Provide information, warnings, and resources to help users understand and avoid engaging with loan sharks • Share relevant data and collaborate with authorities to investigate and combat illegal lending practices | • Ue the ratio to identify the presence of loan shark institutions and mitigate associated risks • Ensure adherence to regulatory requirements, report any unusual patterns, and collaborate with regulatory authorities to address potential issues • Proactively inform users about the risks associated with loan shark institutions, providing guidance on how to recognize and avoid such entities • Incorporate the ratio into risk assessment models to refine credit scoring and lending decisions by adjusting risk parameters based on the prevalence of loan shark-related events, improving the accuracy of risk assessments | {float, null} |
| median_depository_credit _{t0_days}_{t1_days} | The median depository credit transaction observed from the users' SMS. The median is calculated over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Create customized savings plans for users based on their depository credit transaction history. • Introduce investment advisory services for users with significant depository credit transactions • Launch targeted marketing campaigns for specific user segments based on depository credit behavior | • Offer personalized financial planning services based on users' depository credit transactions • Gain insights into saving patterns, recommend investment opportunities, and assist users in achieving their financial goals. | {float, null} |
| median_depository_debit _{t0_days}_{t1_days} | The median depository debit transaction observed from the users' SMS. The median is calculated over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Create customized savings plans for users based on their depository debit transaction history. • Introduce investment advisory services for users with significant depository debit transactions • Launch targeted marketing campaigns for specific user segments based on depository debit behavior | • Offer personalized financial planning services based on users' depository debit transactions • Gain insights into saving patterns, recommend investment opportunities, and assist users in achieving their financial goals. | {float, null} |
| median_end_of_day _depository_balance _{t0_days}_{t1_days} | Median of daily total balance held in all known depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement dynamic savings goals based on the median daily total balance by assisting users in setting realistic savings goals, offering incentives, and providing guidance on achieving financial objectives • Adjust risk management strategies, offer premium services, or provide tailored financial advice to users with higher financial capacity • Customize marketing messages and promotions for users with varying levels of financial stability and capacity | • Assess the overall financial health of users and provide a comprehensive view of a user's financial well-being • Use this data to refine credit scoring models, enabling a more accurate evaluation of a user's ability to manage and repay credit. • Identify users with fluctuating median balances and develop targeted campaigns to enhance user satisfaction, encourage saving, or provide relevant financial guidance | {float, null} |
| median_end_of_day _loan_balance _{t0_days}_{t1_days} | The time-median of the user’s total loan summed across loan accounts at the end of each day. Balances are observed in SMS from known loan institutions. The time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement dynamic interest rates based on the average daily total loan balance by adjusting interest rates on savings accounts or other interest-bearing products to reflect users' financial behaviors and encourage higher average balances • Offer promotions on certain products or services, providing added value to users with higher average balances • Provide insights and recommendations to help users optimize their financial strategies, such as suggesting investment options or debt management tips | • Segment customers and offer targeted financial products and services where users with higher average balances may be eligible for premium services or personalized offerings • Refine credit scoring models, allowing for a more accurate evaluation of a user's financial capacity • Use the average daily total balance as a metric to design user engagement strategies and loyalty programs such as rewards, incentives, or exclusive benefits to users with consistently high average balances | |
| min_end_of_day _depository_balance _{t0_days}_{t1_days} | The minimum end-of-day (EOD) depository total balances, across all institutions where balances are observed in SMS. Total balances means the sum of EOD balances across all institutions on a daily basis. End-of-day (EOD) means the most recent notification of account balance backward looking from the end of each calendar day. The minimum is taken on the EOD observations over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Adjust overdraft limits to align with users' historical minimum balances, providing a safety net for unexpected expenses • Use minimum EOD balances as part of an early warning system for potential financial challenges by proactively reaching out to users experiencing consistently low minimum balances to offer support, advice, or alternative financial solutions | • Help banks and fintechs in understanding the users' ability to maintain a minimum financial threshold and informs credit decisions • Identify users with consistently low minimum EOD balances for targeted assistance program • Tailor financial products for users with low minimum EOD balances | {float, null} |
| mpesa_average_end_of_day _depository_balance _{t0_days}_{t1_days} | Average of daily total balance held in M-Pesa depository account(s), over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement dynamic interest rates based on the average daily total balance by adjusting interest rates on savings accounts or other interest-bearing products to reflect users' financial behaviors and encourage higher average balances • Offer promotions on certain products or services, providing added value to users with higher average balances • Provide insights and recommendations to help users optimize their financial strategies, such as suggesting investment options or debt management tips | • Segment customers and offer targeted financial products and services where users with higher average balances may be eligible for premium services or personalized offerings • Refine credit scoring models, allowing for a more accurate evaluation of a user's financial capacity • Use the average daily total balance as a metric to design user engagement strategies and loyalty programs such as rewards, incentives, or exclusive benefits to users with consistently high average balances | {float, null} |
| mpesa_median_end_of_day _depository_balance _{t0_days}_{t1_days} | Median of daily total balance held in M-Pesa depository account(s), over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement dynamic savings goals based on the median daily total balance by assisting users in setting realistic savings goals, offering incentives, and providing guidance on achieving financial objectives. • Adjust fee structures for users with varying median balances such as fee waivers, reduced fees, or alternative fee structures for users with different median balance profiles to enhance financial inclusivity | • Use the median daily total balance as a key metric for assessing the overall financial health of users • Refine credit scoring models, enabling a more accurate evaluation of a user's ability to manage and repay credit • Create products such as savings accounts, investment options, or credit products specifically designed for users within certain median balance ranges | {float, null} |
| mpesa_min_end_of_day _depository_balance _{t0_days}_{t1_days} | The minimum end-of-day (EOD) depository total balances, across M-Pesa account(s) where balances are observed in SMS. Total balances means the sum of EOD balances across M-Pesa account(s) on a daily basis. End-of-day (EOD) means the most recent notification of account balance backward looking from the end of each calendar day. The minimum is taken on the EOD observations over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Implement dynamic overdraft limits based on minimum EOD balances by adjusting overdraft limits to align with users' historical minimum balances, providing a safety net for unexpected expenses • Provide insights and recommendations to help users optimize their financial strategies, such as suggesting investment options or debt management tips • Provide added value to users with lower account balances by launching targeted promotions and discounts for users with specific minimum EOD balances | • Use the minimum EOD balances as a risk management metric for assessing creditworthiness as it can help in understanding users' ability to maintain a minimum financial threshold and informs credit decisions • Identify users with consistently low minimum EOD balances for targeted assistance programs • Create products such as fee structures, overdraft protection, or savings plans specifically designed to meet the needs of users with lower account balances | {float, null} |
| mpesa_net_cash_flow _{t0_days}_{t1_days} | Difference between credit and debit transactions across M-Pesa depository account(s), over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Evaluate the stability of users' income by analyzing the credit-debit transaction difference • Implement dynamic adjustments to credit limits based on spending and income behaviors by assessing users' capacity for additional credit and offer credit limit adjustments accordingly • Segment users for targeted marketing efforts based on their credit-debit transaction difference and customize marketing messages and promotions for users with varying spending and income patterns | • Use the difference between credit and debit transactions to analyze users' spending and income pattern as this helps in understanding how users manage their finances and can inform targeted financial advice or product recommendations • Provide budgeting assistance based on the credit-debit transaction difference by offering users insights into their spending habits, help set budget goals, and suggest ways to optimize their financial management | {float, null} |
| mpesa_stdev_end_of_day _depository_balance _{t0_days}_{t1_days} | The standard deviation of end-of-day (EOD) depository total balances, across M-Pesa account(s) where balances are observed in SMS. Total balances means the sum of EOD balances across M-Pesa account(s) on a daily basis. End-of-day (EOD) means the most recent notification of account balance backward looking from the end of each calendar day. The standard deviation is calculated on the EOD observations over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Offer tailored savings plans based on users' EOD balance volatility • Offer insights, budgeting tips, and tools to help users manage their finances more effectively during periods of fluctuation • Design targeted campaigns or promotions to engage users during times of financial stability or encourage financial planning during periods of volatility • Refine credit assessments by considering users' historical balance volatility, leading to more accurate creditworthiness evaluations | • Use the standard deviation of EOD balances as a risk management metric to assess users' financial stability where higher standard deviations may indicate financial volatility, helping institutions tailor risk mitigation strategies • Implement dynamic adjustments to credit limits based on variations in EOD balances, assessing users' financial fluctuations and adjust credit limits accordingly, optimizing credit risk management | {float, null} |
| mpesa_sum_of_credits _{t0_days}_{t1_days} | Total of credit transactions across M-Pesa depository account(s), over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Identify users with consistent income patterns and offer financial planning support or customized products • Implement targeted marketing campaigns for users with specific total credit transaction patterns, customizing promotions or offers for products related to income, such as investment options or insurance • Optimize credit limits based on users' total credit transactions and dynamically adjust credit limits, ensuring that users have access to appropriate levels of credit based on their income sources. | • Helps analyze the sources of users' income, enabling targeted product recommendations or financial advice • Evaluate users' capacity to generate income, potentially refining credit scoring models for more accurate lending decisions • Assess users' overall income stability and adjust credit limits accordingly basd on the total of credit transactions, optimizing credit risk management | {float, null} |
| mpesa_sum_of_debits _{t0_days}_{t1_days} | Total of debit transactions across M-Pesa depository account(s), over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement dynamic spending limits based on the total of debit transaction and assess users' overall spending behavior and adjust spending limits accordingly, optimizing risk management • Provide recommendations for savings products or investment options that align with users' financial behaviors and goals • Implement targeted marketing campaigns for users with specific total debit transaction patterns such as promotions or products related to spending, such as cashback rewards or discounts | • Helps analyze users' spending patterns, enabling targeted product recommendations or financial advice • Offer personalized products such as budgeting tools, savings plans, or credit products that align with users' spending patterns | {float, null} |
| mpesa_sum_of_p2p_credits _{t0_days}_{t1_days} | Total sum of M-Pesa credit transactions that were conducted from peer to peer, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Offer promotions on certain products or services, providing added value to users with higher M-Pesa credit transactions conducted from peer to peer. • Provide insights and recommendations to help users optimize their financial strategies, such as suggesting investment options or debt management tips | • Segment customers and offer targeted financial products and services where users with higher M-Pesa credit transactions conducted from peer to peer may be eligible for premium services or personalized offerings • Refine credit scoring models, allowing for a more accurate evaluation of a user's financial capacity • Understanding a user’s M-Pesa credit transactions conducted from peer to peer. can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| mpesa_sum_of_p2p_debits _{t0_days}_{t1_days} | Total sum of M-Pesa debit transactions that were conducted from peer to peer, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Offer promotions on certain products or services, providing added value to users with higher debit M-Pesa debit transactions conducted from peer to peer. • Provide insights and recommendations to help users optimize their financial strategies, such as suggesting investment options or debt management tips | • Segment customers and offer targeted financial products and services where users with higher debit M-Pesa debit transactions conducted from peer to peer may be eligible for premium services or personalized offerings • Refine credit scoring models, allowing for a more accurate evaluation of a user's financial capacity • Understanding a user’s M-Pesa debit transactions conducted from peer to peer. can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| mpesa_transaction_count_credit _fraction_{t0_days}_{t1_days} | The fraction of M-Pesa transactions that are credits transactions, relative to total count of transactions (credits and debits inclusive). | • Create customized savings plans for users based on their M-Pesa transactions credit count fraction. • Introduce investment advisory services for users with significant M-Pesa transactions credit count fraction. • Launch targeted marketing campaigns for specific user segments based on M-Pesa transactions credit count fraction. | • Offer personalized financial planning services based on users' M-Pesa transactions credit count fraction. • Gain insights into saving patterns, recommend investment opportunities, and assist users in achieving their financial goals. | {float, null} |
| net_cash_flow _{t0_days}_{t1_days} | Difference between credit and debit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Design promotions or offers that resonate with specific spending and saving behaviors, enhancing user engagement • Incorporate the difference between credit and debit transactions into credit scoring models and enhance the accuracy of credit assessments by considering users' transactional behaviors for more informed lending decisions • Incorporate the difference between credit and debit transactions into credit scoring models and enhance the accuracy of credit assessments by considering users' transactional behaviors for more informed lending decisions | • Gain a comprehensive understanding of users' financial behavior by examining the difference between credit and debit transactions • Provide intelligent budgeting assistance based on the detailed analysis of credit and debit transaction differences • Optimize spending limits dynamically based on a nuanced understanding of users' overall spending behavior • Establish an early warning system to detect potential financial challenges by monitoring changes in credit-debit transaction patterns | {float, null} |
| primary_currency | Returns the primary currency based on the user's primitives data. | • Automate currency conversion processes based on the identified primary currency, providing users with real-time and accurate exchange rates • Customize user interfaces based on the primary currency, displaying financial information in a format that is familiar and relevant to the user • Tailor product recommendations, such as savings accounts or investment options, based on the user's identified primary currency and financial goals • Facilitate international transactions by recognizing the primary currency, allowing for seamless cross-border payments and reducing currency-related complexities • Strengthen user authentication processes by incorporating the identification of the primary currency, adding an additional layer of security and verification | • Enhance the user experience by automatically determining and presenting the primary currency based on the user's primitives data • Improve transaction processing efficiency by automatically identifying the primary currency, reducing errors and providing a seamless financial experience • Facilitate cross-border financial planning by recognizing the user's primary currency, enabling more accurate budgeting and investment decisions • Mitigate currency risk by accurately determining the user's primary currency, helping users make informed decisions about currency fluctuations | {str, null} |
| slope_end_of_day _depository_balance_0_90 | Rate of change of daily total balance held in all depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement dynamic adjustments to spending limits based on the rate of change of daily total balances, ensuring users have appropriate limits aligned with their financial behaviors • Optimize credit limits based on the rate of change of daily total balances, ensuring that users have access to appropriate levels of credit aligned with their evolving financial situations • Develop predictive user engagement strategies by analyzing trends in the rate of change of daily total balances, allowing for targeted campaigns or promotions to enhance user satisfaction | • Provide a dynamic assessment of users' financial health by analyzing the rate of change of daily total balances, offering insights into their financial stability and habits • Enable real-time budgeting insights by calculating the rate of change of daily total balances, helping users make informed decisions based on their recent financial activities • Facilitate early detection of potential financial strain by analyzing the rate of change of daily total balances, allowing for proactive financial support or guidance • Enhance credit scoring models by considering the rate of change of daily total balances, providing a more comprehensive view of users' financial behaviors for improved credit assessments | {float, null} |
| slope_end_of_day _loan_balance_0_90 | Rate of change of daily total balance held in all loan accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Establish an early warning system for potential loan delinquency by monitoring abnormal patterns in the rate of change of daily total balances, allowing for timely borrower engagement • Implement loan restructuring initiatives based on the rate of change of daily total balances, providing flexible repayment options to borrowers facing financial challenges • Develop predictive models for loan default analysis by considering trends in the rate of change of daily total balances, enabling more accurate risk assessments • Tailor loan products based on the rate of change of daily total balances, offering products that align with borrowers' repayment behavior and financial capacity | • Monitor the performance of loans by analyzing the rate of change of daily total balances in loan accounts, providing insights into borrowers' repayment behaviors • Assess and mitigate risks associated with loans by tracking the rate of change of daily total balances, identifying potential issues and implementing proactive measures • Offer real-time indicators of borrowers' financial health through the rate of change of daily total balances in loan accounts, facilitating timely interventions and support • Optimize loan portfolio strategies by leveraging insights from the rate of change of daily total balances in loan accounts, refining lending practices for better overall portfolio performance | {float, null} |
| stdev_end_of_day _depository_balance _{t0_days}_{t1_days} | The standard deviation of end-of-day (EOD) depository total balances, across all institutions where balances are observed in SMS. Total balances means the sum of EOD balances across all institutions on a daily basis. End-of-day (EOD) means the most recent notification of account balance backward looking from the end of each calendar day. The standard deviation is calculated on the EOD observations over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement real-time risk mitigation strategies by monitoring the standard deviation of EOD depository total balances, allowing financial institutions to proactively manage potential financial risks • Refine credit scoring models by incorporating insights from the standard deviation of EOD balances, improving the accuracy of credit assessments and lending decisions • Provide customized credit limit recommendations based on the observed standard deviation of EOD balances, ensuring that users have appropriate credit limits aligned with their financial behavior • Set up fraud alert triggers based on abnormal fluctuations in the standard deviation of EOD balances, enabling rapid response to potential fraudulent activities | • Provide a comprehensive risk management tool by calculating the standard deviation of EOD depository total balances, enabling financial institutions to assess volatility in users' financial positions • Support dynamic adjustments to credit limits based on the standard deviation of EOD balances, allowing for real-time optimization of credit risk management • Offer users personalized insights into their financial stability by analyzing the standard deviation of EOD depository total balances, providing actionable information for improved financial planning • Develop more targeted customer engagement strategies by considering the standard deviation of EOD balances, enabling financial institutions to offer relevant products and services based on users' financial behaviors | {float, null} |
| sum_of_airtime_credits _{t0_days}_{t1_days} | Total of credit transactions across airtime accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Implement targeted marketing campaigns for communication products by leveraging insights from the total of credit transactions, ensuring that promotions are relevant to users' communication habits | • Gain valuable insights into users' behavior by calculating the total of credit transactions across airtime accounts, allowing for a deeper understanding of their communication and spending patterns • Improve customer engagement strategies by analyzing the total of credit transactions in airtime accounts, allowing for targeted promotions, discounts, or loyalty programs • Tailor airtime offers based on the total of credit transactions, providing users with personalized promotions that align with their communication needs and usage. • Inform user-centric product development by considering the total of credit transactions, helping in the creation of new services or features that meet users' communication and financial needs • Optimize revenue streams by understanding the total of credit transactions in airtime accounts, allowing for strategic pricing, bundling, and monetization of communication services | {float, null} |
| sum_of_atm_credits _{t0_days}_{t1_days} | Total of ATM (automatic teller machine) credit deposits across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Drive financial literacy initiatives by leveraging insights from the total of atm credit deposits, allowing banks and fintechs to educate users in African markets on effective financial management practices • Implement localized loyalty programs based on the total of atm credit deposits, allowing banks and fintechs to reward users for their airtime spending and encourage continued engagement with their services • Optimize mobile wallet services by incorporating the total of atm credit deposits, enabling banks and fintechs to enhance the functionality and usability of their digital wallets for seamless airtime transactions | • Facilitate seamless integration with mobile money services by leveraging insights from the total of atm credit deposits, allowing banks and fintechs to offer complementary financial services and enhance user experiences • Unlock digital lending opportunities by incorporating the total of atm credit deposits into credit scoring models, enabling banks and fintechs to extend micro-loans or credit to users based on their airtime usage | {float, null} |
| sum_of_atm_debits _{t0_days}_{t1_days} | Total of ATM (automatic teller machine) debit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Enhance credit scoring models by incorporating insights from the total of ATM (automatic teller machine) debit transactions, providing a more accurate assessment of users' creditworthiness for improved lending decisions • Optimize savings and investment recommendations by considering the total of ATM (automatic teller machine) debit transactions, helping users make informed decisions aligned with their financial goals | • Enable comprehensive analysis of users' financial behavior by calculating the total of ATM (automatic teller machine) debit transactions across depository accounts, providing insights into spending patterns and financial habits • Customize credit and loan offerings based on the total of ATM (automatic teller machine) debit transactions, allowing banks and fintechs to tailor financial products to users' spending behaviors and repayment capacities • Provide budgeting assistance and support financial planning by analyzing the total of ATM (automatic teller machine) debit transactions, helping users manage their finances effectively and achieve their financial goals | {float, null} |
| sum_of_credits _{t0_days}_{t1_days} | Total of credit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Support microfinance decision-making by analyzing the total of credit transactions, enabling banks and fintechs to assess the creditworthiness of users in regions with limited traditional banking infrastructure • Drive financial literacy initiatives by leveraging insights from the total of credit transactions, allowing banks and fintechs to educate users in African markets on effective financial management practices • Implement localized loyalty programs based on the total of credit transactions, allowing banks and fintechs to reward users for their airtime spending and encourage continued engagement with their services • Optimize mobile wallet services by incorporating the total of credit transactions, enabling banks and fintechs to enhance the functionality and usability of their digital wallets for seamless airtime transactions | • Enhance financial inclusion efforts by analyzing the total of credit transactions across airtime accounts, providing insights into users' communication and transaction behaviors in regions with limited banking infrastructure. • Facilitate seamless integration with mobile money services by leveraging insights from the total of credit transactions, allowing banks and fintechs to offer complementary financial services and enhance user experiences • Drive localized customer engagement strategies by understanding the total of credit transactions, enabling targeted campaigns that resonate with the unique communication habits of users in African markets • Unlock digital lending opportunities by incorporating the total of credit transactions into credit scoring models, enabling banks and fintechs to extend micro-loans or credit to users based on their airtime usage • Optimize cross-border remittance services by considering the total of credit transactions, allowing banks and fintechs to tailor remittance products to users' communication needs and spending behaviors | {float, null} |
| sum_of_debits _{t0_days}_{t1_days} | Total of debit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Enhance credit scoring models by incorporating insights from the total of debit transactions, providing a more accurate assessment of users' creditworthiness for improved lending decisions • Optimize savings and investment recommendations by considering the total of debit transactions, helping users make informed decisions aligned with their financial goals • Establish an early warning system for potential financial challenges by monitoring changes in the total of debit transactions, allowing proactive engagement with users facing financial strain • Improve operational efficiency by understanding the total of debit transactions, optimizing internal processes related to account management, transaction processing, and customer support | • Enable comprehensive analysis of users' financial behavior by calculating the total of debit transactions across depository accounts, providing insights into spending patterns and financial habits • Facilitate seamless integration with digital banking services by leveraging insights from the total of debit transactions, enhancing user experiences and offering a more connected financial ecosystem • Customize credit and loan offerings based on the total of debit transactions, allowing banks and fintechs to tailor financial products to users' spending behaviors and repayment capacities • Provide budgeting assistance and support financial planning by analyzing the total of debit transactions, helping users manage their finances effectively and achieve their financial goals | {float, null} |
| sum_of_depository _balances_latest | The sum of the latest balances for depository accounts, across all institutions where balances are observed in SMS. Latest means the most recent notification of account balance backward looking from the prediction date. | • Implement dynamic adjustments to credit limits based on the sum of the latest balances, ensuring users have access to appropriate credit facilities aligned with their financial positions • Mitigate credit risk by incorporating insights from the sum of the latest balances into risk management strategies, allowing for proactive measures to address potential financial challenges • Launch targeted savings and investment campaigns based on the sum of the latest balances, encouraging users to explore wealth-building opportunities aligned with their financial capacity • Drive enhanced customer engagement by leveraging insights from the sum of the latest balances, allowing banks and fintechs to engage users with timely and relevant financial offers and updates | • Provide users with a holistic financial snapshot by calculating the sum of the latest balances for depository accounts, enabling a comprehensive view of their overall financial health • Customize financial products based on the sum of the latest balances, allowing banks and fintechs to tailor services and offers to meet the specific needs and preferences of users • Improve credit scoring models by incorporating the sum of the latest balances, providing a more accurate assessment of users' financial stability and creditworthiness • Implement targeted marketing campaigns for savings products based on the sum of the latest balances, encouraging users to explore and engage with savings and investment options | {float, null} |
| sum_of_depository _transactions_between _user_accounts _{t0_days}_{t1_days} | Total sum of transaction amounts potentially taking place between depository accounts associated with the given user, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Customize product offerings based on the potential transaction amounts, allowing banks and fintechs to create tailored solutions that address users' specific transaction needs. • Identify enhanced cross-selling opportunities by analyzing potential transaction amounts, enabling targeted promotions or offerings that align with users' anticipated financial activities • Develop integrated digital payment solutions by understanding potential transaction amounts between depository accounts, providing users with convenient and efficient payment options • Enable real-time budget adjustments for users by considering potential transaction amounts, allowing for dynamic budgeting based on anticipated financial activities • Optimize digital wallet services by incorporating insights from potential transaction amounts, enhancing the functionality and user experience of digital wallets for seamless fund transfers | • Enhance visibility into potential inter-account transactions by calculating the total sum of transaction amounts, providing a comprehensive view of funds movement within a user's associated depository accounts • Foster an integrated financial ecosystem by understanding the total sum of transaction amounts between depository accounts, enabling banks and fintechs to offer seamless and connected financial services • Bundle products and services based on the potential transaction amounts between depository accounts, creating user-centric packages that align with diverse financial needs • Provide personalized financial recommendations based on the total sum of potential transaction amounts, assisting users in optimizing their financial behaviors and achieving their financial goals | {float, null} |
| sum_of_[fastfood_events _categories]_events _{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending through food spends, where food spends are broken down into general kfc,java, chicken inn, creamy inn, pizza inn, and galitos. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer service and build stronger relationships with customers, thereby increasing customer retention by understanding customers’ spending habits and preferences on food spends. • Use data to offer personalized services such as budgeting and financial planning tools that track food events and compare pricing and packages from different catering or food precessing services, helping them make cost-conscious decisions and manage their finances better. • Improve data quality and reliability by using SMS data as a source of truth for verifying food spends of customers. | • These features may be indicative of individuals who frequently dine out, possibly suggesting a personal or professional interest in the food and drink industry. • Understanding a user’s activity related to food events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[financial_ spending_ categories] _debits_{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending through financial spends, where financial spends is broken down into general financial spending, insurance, investment, SACCO and service spends. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance customer engagement and retention by offering tailored product recommendations for financial spends such as insurance, investment and savings services, helping them make cost-conscious decisions and manage their finances better. • Improve data quality and reliability by using SMS data as a source of truth for verifying financial spends of customers. | • These features may be indicative of an individual’s interest in financial growth through investments and corporate savings. It may also point to individuals owning high-valued properties that necessitate insurance. • Understanding a user’s activity related to financial events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[food_spending _categories]_debits _{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending through food spends, where food spends are broken down into general food spending, bar and restaurant, catering, food processor and distributor, food store, wine and spirit spends. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer service and build stronger relationships with customers, thereby increasing customer retention by understanding customers’ spending habits and preferences on food spends. • Use data to offer personalized services such as budgeting and financial planning tools that track food events and compare pricing and packages from different catering or food precessing services, helping them make cost-conscious decisions and manage their finances better. • Improve data quality and reliability by using SMS data as a source of truth for verifying food spends of customers. | • These features may be indicative of individuals who frequently dine out, possibly suggesting a personal or professional interest in the food and drink industry. • Understanding a user’s activity related to food events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_golf_clubs_debits _{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending golf events The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days | Enhance customer engagement and retention by offering tailored recommendations on golfing events • Improve data quality and reliability by using SMS data as a source of truth for verifying spends of customers. | • These features may be indicative of the amount spent on golfclubs-related services, potentially highlighting individuals who frequent this area for targeted marketing and partnerships • Understanding a user’s activity related to golf events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[health_spending _categories]_debits _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through health spends, where health spends are broken down into general health, hospital, pharmaceutical, healthcare and specialist spends. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance customer engagement and retention by offering tailored recommendations on healthy living, tips and HMOs. • Improve data quality and reliability by using SMS data as a source of truth for verifying health spends of customers. | • These features may be indicative of the amount spent on health-related services, potentially highlighting individuals facing health challenges who could be targeted for health product campaigns. • Understanding a user’s activity related to health events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_loan_debits _{t0_days}_{t1_days} | Total of loan debit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Understand users' lending patterns across differnt lenders for proper segmentation for marketting and product offerings • Enhance credit scoring models by incorporating insights from the total of loan debit transactions, providing a more accurate assessment of users' creditworthiness for improved lending decisions | • Enable comprehensive analysis of users' financial behavior by calculating the total of loan debit transactions across depository accounts, providing insights into spending patterns and financial habits • Customize credit and loan offerings based on the total of loan debit transactions, allowing banks and fintechs to tailor financial products to users' spending behaviors and repayment capacities | {float, null} |
| sum_of_loan_repayments _{t0_days}_{t1_days} | The sum of repayment amounts (credit transactions) for all loans from known lenders, where a repayment amount is observed in the SMS indicating the repayment. The summation is computed for all repayment events occurring over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Foster customer satisfaction by offering personalized solutions, such as flexible repayment schedules or targeted promotions, based on observed repayment amounts • Improve risk assessment by identifying customers with consistent repayment behavior and offering them lower-risk financial products or increased credit limits • Enhance cross-selling opportunities and attract responsible borrowers by Leverage insights from repayment amounts such as offering exclusive loan terms, interest rates, or rewards to customers with a history of timely repayments | • Provide deeper understanding of customers' repayment behavior and financial responsibility • Optimize pricing and risk strategies by leveraging the sum of repayment amounts • Segmenting customers according to their repayment behavior and tailor products and services based on customers' repayment history | {float, null} |
| sum_of_mpesa_buygoods_debits _{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending through M-Pesa BuyGoods. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer satisfaction and loyalty by offering customers access to exclusive shopping platform events, such as flash sales, early access events, and member-only discounts, rewarding their loyalty and encouraging repeat purchases. • Improve data quality and reliability by using SMS data as a source of truth for verifying buygoods events of customers. | • This feature may be indicative of an individual’s general spending power on goods and products, reflecting their economic capability to purchase various items. • Understanding a user’s activity related to M-Pesa BuyGoods can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_mpesa_paybill_debits _{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through M-Pesa Paybill. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer satisfaction by recommending platforms that provide ease of payment for bills • Improve data quality and reliability by using SMS data as a source of truth for verifying paybill events of customers. | • This feature may be indicative of an individual’s general spending power on bills and services, reflecting their economic capability to service various bills. • Understanding a user’s activity related to M-Pesa Paybill can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_non_mpesa_credits _{t0_days}_{t1_days} | Total of credit transactions across non-mpesa depository accounts. The summation is computed for all non-mpesa credit events occurring over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Identify users with consistent income patterns in their non-mpesa accounts and offer financial planning support or customized products • Implement targeted marketing campaigns for users with specific total credit transaction patterns, customizing promotions or offers for products related to income, such as investment options or insurance. • Optimize credit limits based on users' total non-mpesa credit transactions and dynamically adjust credit limits, ensuring that users have access to appropriate levels of credit based on their income sources. | • Helps analyze the sources of users' non-mpesa income, enabling targeted product recommendations or financial advice • Evaluate users' capacity to generate income, potentially refining credit scoring models for more accurate lending decisions • Assess users' overall income stability and adjust credit limits accordingly basd on the total of credit transactions, optimizing credit risk management. | {float, null} |
| sum_of_non_mpesa_debits _{t0_days}_{t1_days} | Total of debit transactions across non-mpesa depository accounts. The summation is computed for all non-mpesa debit events occurring over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Identify users with consistent spending patterns in their non-mpesa accounts and offer financial planning support or customized products • Implement targeted marketing campaigns for users with specific total debit transaction patterns, customizing promotions or offers for products related to income, such as investment options or insurance. • Optimize credit limits based on users' total non-mpesa debit transactions and dynamically adjust credit limits, ensuring that users have access to appropriate levels of credit based on their income sources. | • Helps analyze the sources of users' non-mpesa spending, enabling targeted product recommendations or financial advice • Evaluate users' non-mpesa spending patterns, potentially refining credit scoring models for more accurate lending decisions • Assess users' overall spending variability and adjust credit limits accordingly basd on the total of credit transactions, optimizing credit risk management. | {float, null} |
| sum_of_[public_institutions _spending_categories] _debits_{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through public institutions, where public institutions are broken down into general public institutions, charity, church, government and school spends. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | •Increase customer engagement and retention by personalized recommendations for public institution such as charity and churches based on their past donations and interests • Improve customer satisfaction and loyalty by offering loans for school fees payment at the needed time • Improve data quality and reliability by using SMS data as a source of truth for verifying public institution spends of customers. | • These features may be indicative of individuals belonging to a religious community and inclined towards charitable giving. It may also suggest individuals with dependents in educational institutions or those pursuing education themselves. • Understanding a user’s activity related to public institution events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_pos_credits _{t0_days}_{t1_days} | Total of POS (point of sale) credit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Drive financial literacy initiatives by leveraging insights from the total of POS (point of sale) credit transactions, allowing banks and fintechs to educate users in African markets on effective financial management practices • Implement localized loyalty programs based on the total of POS (point of sale) credit transactions, allowing banks and fintechs to reward users for their airtime spending and encourage continued engagement with their services • Optimize mobile wallet services by incorporating the total of POS (point of sale) credit transactions, enabling banks and fintechs to enhance the functionality and usability of their digital wallets for seamless airtime transactions | • Facilitate seamless integration with mobile money services by leveraging insights from the total of POS (point of sale) credit transactions, allowing banks and fintechs to offer complementary financial services and enhance user experiences • Unlock digital lending opportunities by incorporating the total of POS (point of sale) credit transactions into credit scoring models, enabling banks and fintechs to extend micro-loans or credit to users based on their airtime usage | {float, null} |
| sum_of_pos_debits _{t0_days}_{t1_days} | Total of POS (point of sale) debit transactions across depository accounts, over the previous t0 to t1 days, where the time windows are 0-30, 31-90, or 0-90 days history prior to the prediction date. | • Enhance credit scoring models by incorporating insights from the total of POS (point of sale) debit transactions, providing a more accurate assessment of users' creditworthiness for improved lending decisions • Optimize savings and investment recommendations by considering the total of POS (point of sale) debit transactions, helping users make informed decisions aligned with their financial goals | • Enable comprehensive analysis of users' financial behavior by calculating the total of POS (point of sale) debit transactions across depository accounts, providing insights into spending patterns and financial habits • Customize credit and loan offerings based on the total of POS (point of sale) debit transactions, allowing banks and fintechs to tailor financial products to users' spending behaviors and repayment capacities • Provide budgeting assistance and support financial planning by analyzing the total of POS (point of sale) debit transactions, helping users manage their finances effectively and achieve their financial goals | {float, null} |
| sum_of_salary_credits _{t0_days}_{t1_days} | The number of SMS received indicating that the user received income through salary. The credits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Use the data to build predictive models that forecast the likelihood of a user defaulting on a loan based on estimated monthly income • Improve data quality and reliability by using SMS data as a source of truth for verifying salary income of customers. | • These features may be indicative of the inflow an individual receives on a monthly basis, serving as useful data for estimating income. • Understanding a user’s activity related to Salary income can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[shopping_spending _categories] _debits_{t0_days}_{t1_days} | The number of SMS received indicating that the user engaged in spending through shopping events, where shopping events are broken down into general shopping spending, clothing and fashion, ecommerce platform, entertainment and subscription, professional service, retail store, supermarket, naivas, quickmart, carrefour, wholesale and distribution . The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Improve customer satisfaction and loyalty by offering customers access to exclusive shopping platform events, such as flash sales, early access events, and member-only discounts, rewarding their loyalty and encouraging repeat purchases. • Enhance customer engagement and retention by offering tailored product recommendations, enhancing their shopping experience and increasing the likelihood of purchase • Improve data quality and reliability by using SMS data as a source of truth for verifying shopping spends of customers. | • These features may be indicative of an individual’s spending capacity across a broad spectrum of goods and services, offering insights into their overall financial behavior and preferences. • Understanding a user’s activity related to shopping events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[transportation_and _travel_spending_categories] _debits_{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending through transportation and travel, where transportation and travel spends are broken down into general transportation and travel spending, airfare, hotel, logistics, parking and petrol station spends. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Increase customer loyalty and retention by offering incentives on transportation and travel services, such as discounts, or redeemable purchase points on airfare, hotel etc. • Offer personalized services such as airfare events comparison across different airlines & dates or offer lodging options to customers who frequently engage in travel for smoother commute. • Improve data quality and reliability by using SMS data as a source of truth for verifying transportation and travel spends of customers | • These features may be indicative of individuals who travel, commute, or vacation regularly, as well as those owning vehicles, suggesting a certain lifestyle or mobility requirement. • Understanding a user’s activity related to transportation and travel events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[utilities_spending _categories]_debits _{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending through utilities, where utilities are broken down into general utilities spending, airtime, energy, kplc, housing, internet, and water bill spends. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | • Enhance customer engagement and retention by enabling customers to track their spend and usage patterns across different utility events inorder to manage consumption. • Improve customer satisfaction by recommending platform that provide ease of payment for utility bills • Improve data quality and reliability by using SMS data as a source of truth for verifying utility events of customers. | These features may be indicative of individuals who earn regularly and possess sufficient resources to cover short-term bills monthly or weekly. Such individuals are likely to be relatively stable financially, as evidenced by their consistent bill payments. • Understanding a user’s activity related to utilities events can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
| sum_of_[vendors\possystems _categories]_debits _{t0_days}_{t1_days} | The sum of SMS received indicating that the user engaged in spending and trascating through pos systems broken down into gallantchoices, japasupplies, startbase enterprises, merryland stores, sweetbargain traders, kiodistributors, kopokopo, paystack, pesapal, sokowatch. The debits are summed over a period of t0 to t1 days history prior to the prediction date, where the time windows are 0-30, 31-90, or 0-90 days. | Identify the vendors used by small business owners • Get insights on the type of business one is engaging in. • Improve decision making when loaning to small business owners by understanding their patterns better. | These features may be indicative of individuals who buy from vendors . These enable one to get a good idea of the client and personalize products such as loans or products related to small business owners • Understanding a user’s business activity can help tailor marketing strategies and develop products that cater to this specific customer segment. | {float, null} |
The {feature}_was_imputed feature equal to 1 indicates that the feature value had a missing value imputed. | Imputation is performed for features that can reasonably have a missing value replaced. .............................................. | • Conduct customer segmentation analysis with imputed features, allowing businesses to better understand and target different customer groups based on more complete datasets • Improve credit scoring and risk assessment models by imputing missing values in relevant features, providing more accurate predictions of individuals' creditworthiness • Optimize supply chain operations by imputing missing values in relevant features, allowing for more accurate demand forecasting and inventory management ..................................................... | • Enhance data completeness and accuracy by performing imputation for features with missing values, ensuring that datasets used for analysis and modeling are more robust and reliable • Improve predictive modeling outcomes by imputing missing values in features, allowing for more accurate and representative models with reduced bias and improved performance • Provide enhanced decision-making support by imputing missing values in features, enabling stakeholders to make informed decisions based on comprehensive and well-structured data. • Enhance time and resource efficiency in data analysis by automating the imputation process for features with missing values, allowing data scientists and analysts to focus on deriving insights rather than handling missing data • Strengthen data quality assurance practices by systematically imputing missing values in features, contributing to the overall reliability and trustworthiness of the data used in various business processes ............................................. | {boolean} ..................... |