System and method for issuing a loan to a consumer determined to be creditworthy and with bad debt forecast

Abstract

A system and method determines the creditworthiness of a consumer and issues a loan and generates a behavioral profile for that consumer. An initial set of data is acquired from the consumer that includes non-identification attributes without obtaining a full name, a credit card number, a passport number, or a government issued ID number that allows identification of the consumer. A user ID number matches the initial set of data to a physical user in a transaction database. A credit score based on the average credit among a plurality of user profiles is matched to determine a maximum credit for the consumer. A machine learning model may be applied to stored consumer loan data to determine when the consumer requires an increase in the maximum allowed credit and the risk involved with increasing the maximum allowed credit.

That which is claimed is: 
     
         1 . A method of determining the creditworthiness and issuing a micro- or nano loan to a consumer and forecasting a bad debt probability of that consumer, comprising:
 a consumer selecting and connecting a communications device of the consumer via a communications network to a server system having a communications module, a controller, a database connected thereto, and an application programming interface (API) operative to allow interaction between the server and the communications device,   in response to the consumer's selecting and connecting to the server system, initiating via the API a user interface on a display of the communications device, the user interface displaying a first menu item as a button selection on a portion of the display for requesting a micro- or nano loan, wherein the consumer selects the first menu item and initiates an API call as a request for a micro- or nano loan;   in response to the consumer selecting the first menu item and initiating the loan request, the server system extracts N attributes about the consumer from external public data sources, wherein the N attributes have limited or no personal identification data and confidential information about the consumer and comprise limited or anonymous consumer transaction data extracted from transactional platforms and data extracted from one or more of a) gender, b) age, c) cellular operator, phone model, and usage, d) consumer geolocation, e) home values by geolocation, f) average income by: geolocation, gender and age groups, g) education by: geolocation and gender, h) public transport options by geolocation, i) social media activities by: geolocation, gender and age groups, j) infrastructure and services available by geolocation, and k) criminal records by geolocation;   processing the N attributes at the server system by applying a features construction model and transforming the N attributes into a user attribute string associated with the consumer;   matching the user attribute string associated with the consumer with user attribute strings stored within the database and associated with other consumers, wherein a match to another user attribute string stored within the database is indicative of the micro- or nano loan amount as a maximum credit limit that is loaned to the consumer initially requesting the loan;   transmitting to the communications device a loan approval code, and in response to receiving the loan approval code at the communications device, displaying on the user interface a second menu item as button selections for confirming and selecting a micro- or nano loan amount up to the maximum credit allowed for the consumer and how the loan is to be dispersed as either crediting an electronic wallet of the consumer or paying all or part of a bill associated with an account of the consumer in the value of the loan;   in response to the consumer selecting the second menu item and confirming and selecting a micro- or nano loan amount up to the maximum credit allowed for the consumer and how the loan is to be dispersed, the server system credits the electronic wallet of the consumer or pays all or part of a bill associated with an account of the consumer in the value of the loan based upon the consumer's selection at the second menu item, wherein the micro- or nano loan is approved on an average in under 20 seconds and with no more than three selections entered by the consumer on their communications device;   generating a user ID associated with the user attribute string of the consumer and storing the user ID and user attribute string within the database;   acquiring additional attributes linked to transactions made by the consumer over time;   linking the additional attributes to the consumer's user attribute string stored in the database; and   applying a bad debt prediction model to the additional attributes and user attribute string to generate a bad debt prediction for the consumer as a numerical indicia, and if the numerical indicia is below a threshold value, raising the credit limit for the consumer.   
     
     
         2 . The method according to  claim 1 , wherein the database comprises a first database configured to store user attribute strings of other consumers and a plurality of different pre-approved loan amounts generated from the user attribute strings of the other consumers to which a match is made to determine the micro- or nano loan amount that is loaned to the consumer requesting the loan. 
     
     
         3 . The method according to  claim 2 , wherein the first database comprises a relational database. 
     
     
         4 . The method according to  claim 1 , wherein the communications network comprises a wireless communications network, and the communications device comprises a mobile wireless communications device. 
     
     
         5 . The method according to  claim 1 , wherein the database comprises a second database configured to store the user ID associated with the user attribute string of the consumer that has received the micro- or nanoloan and the additional attributes linked to the consumer over time. 
     
     
         6 . The method according to  claim 5 , wherein the second database comprises a non-relational database. 
     
     
         7 . The method according to  claim 1 , wherein the “N” attributes are extracted without obtaining a full name, a credit card number, a passport number, or a government issued ID number or other data that allows complete identification of the consumer. 
     
     
         8 . The method according to  claim 1 , wherein the additional attributes include data associated with previous purchasing transactions of the consumer over time, wherein the bad debt prediction model comprises a regression model having a moving window that takes into account mean, standard deviation, median, kurtosis and skewness. 
     
     
         9 . The method according to  claim 8 , wherein the server system is configured to input past input/output data to the bad debt prediction model, wherein the past input/output data comprises a vector for the input relating to past consumer loan data and an output relating to a probability between 0 and 1 that indicates whether a consumer will fall into bad debt. 
     
     
         10 . The method according to  claim 9 , wherein a probability greater than 0.6 from the bad debt prediction model is indicative of a high risk that a consumer will fall into bad debt. 
     
     
         11 . The method according to  claim 9 , wherein a target variable outcome from the bad debt prediction model comprises a binary outcome that indicates whether a consumer will be a risk of bad debt within seven days. 
     
     
         12 . The method according to  claim 9 , wherein the server system is configured to collect consumer loan data over a period of six months and classify consumers in two classes as 1) a bad client having a high risk probability of falling into bad debt, and 2) a good client having a low risk probability of falling into bad debt. 
     
     
         13 . The method according to  claim 1 , wherein the server system is configured to generate a behavioral prediction of the consumer and match consumer location and check-ins to at least one of the electronic wallet and the location of the consumer against a known-locations database incorporated within the database and comprising data regarding stores, private locations, public places, and transaction data and correlate periodic location patterns to loan and transactional activities by consumer profile and periodicity;
 loan disbursement patterns;   use of loans;   loan repayments; and   transaction activities.   
     
     
         14 . A system of determining the creditworthiness and issuing a micro- or nano loan to a consumer and forecasting a bad debt probability of that consumer, comprising:
 a communications device of the consumer;   a server system having a communications module, a controller, a database connected thereto, and an application programming interface (API), wherein said API of said server system is operative to allow interaction between the server system and the communications device via a communications network;   in response to the consumer's selecting and connecting to the server system, the server system initiates via the API a user interface on a display of the communications device, the user interface displaying a first menu item as a button selection on a portion of the display for requesting a micro- or nano loan via the first menu item and initiates an API call as a request for a micro- or nano loan;   in response to the consumer selecting the first menu item and initiating the loan request, the server system is configured to extract N attributes about the consumer from external public data sources, wherein the N attributes have limited or no personal identification data and confidential information about the consumer and comprises limited or anonymous consumer transaction data extracted from transactional platforms and data extracted from one or more of a) gender, b) age, c) cellular operator, phone model, and usage, d) consumer geolocation, e) home values by geolocation, f) average income by: geolocation, gender and age groups, g) education by: geolocation and gender, h) public transport options by geolocation, i) social media activities by: geolocation, gender and age groups, j) infrastructure and services available by geolocation, and k) criminal records by geolocation, wherein the server system is configured to:
 process the N attributes at the server system and apply a features construction model and transform the N attributes into a user attribute string associated with the consumer; 
 match the user attribute string associated with the consumer with user attribute strings stored within the database and associated with other consumers, wherein a match to another user attribute string stored within the database is indicative of the micro- or nano loan amount as a maximum credit limit that is loaned to the consumer initially requesting the loan; 
 transmit to the communications device a loan approval code, and in response to receiving the loan approval code at the communications device, the communications device displays on the user interface a second menu item as button selections for confirming and selecting a micro- or nano loan amount up to the maximum credit allowed for the consumer and how the loan is to be dispersed as either crediting an electronic wallet of the consumer or paying all or part of a bill associated with an account of the consumer in the value of the loan; 
   in response to the consumer selecting the second menu item and confirming and selecting a micro- or nano loan amount up to the maximum credit allowed for the consumer and how the loan is to be dispersed, the server system credits the electronic wallet of the consumer or pays all or part of a bill associated with an account of the consumer in the value of the loan based upon the consumer's selection at the second menu item, wherein the micro- or nano loan is approved on an average in under 20 seconds and with no more than three selections entered by the consumer on their communications device;   wherein the server system is configured to:   generate a user ID associated with the user attribute string of the consumer and store the user ID and user attribute string within the database;   acquire additional attributes linked to the transactions made by the consumer over time;   link the additional attributes to the consumer's user attribute string stored in the database; and   apply a bad debt prediction model to the additional attributes and user attribute string to generate a bad debt prediction for the consumer as a numerical indicia, and if the numerical indicia is below a threshold value, the credit limit is raised for the consumer.   
     
     
         15 . The system according to  claim 14 , wherein the database comprises a first database configured to store user attribute strings of other consumers and a plurality of different pre-approved loan amounts generated from the user attribute strings of the other consumers to which a match is made to determine the micro- or nano loan amount that is loaned to the consumer requesting the loan. 
     
     
         16 . The system according to  claim 15 , wherein the first database comprises a relational database. 
     
     
         17 . The system according to  claim 14 , wherein the communications network comprises a wireless communications network, and the communications device comprises a mobile wireless communications device. 
     
     
         18 . The system according to  claim 14 , wherein the database comprises a second database configured to store the user ID associated with the user attribute string of the consumer that has received the micro- or nanoloan and the additional attributes linked to the consumer over time. 
     
     
         19 . The system according to  claim 18 , wherein the second database comprises a non-relational database. 
     
     
         20 . The system according to  claim 14 , wherein the “N” attributes are extracted without obtaining a full name, a credit card number, a passport number, a government issued ID number or other data that allows complete identification of the consumer. 
     
     
         21 . The system according to  claim 14 , wherein the additional attributes include data associated with previous purchasing transactions of the consumer over time, and wherein the bad debt prediction model comprises a regression model having a moving window that takes into account mean, standard deviation, median, kurtosis and skewness. 
     
     
         22 . The system according to  claim 21 , wherein the server system is configured to input past input/output data about the additional attributes to the bad debt prediction model, wherein the past input/output data comprises a vector for the input relating to past consumer loan data and an output relating to a probability between 0 and 1 that indicates whether a consumer will fall into bad debt. 
     
     
         23 . The system according to  claim 22 , wherein a probability greater than 0.6 is indicative of a high risk that a consumer will fall into bad debt. 
     
     
         24 . The system according to  claim 23 , wherein a target variable outcome from the bad debt prediction model comprises a binary outcome that indicates whether a consumer will be a risk of bad debt within seven days. 
     
     
         25 . The system according to  claim 22 , wherein the server system is configured to collect the additional attributes over a period of six months and classify consumers in two classes as 1) a bad client having a high risk probability of falling into bad debt, and 2) a good client having a low risk probability of falling into bad debt. 
     
     
         26 . The system according to  claim 22 , wherein the server system is configured to generate a behavioral prediction of the consumer and match consumer location and check-ins to at least one of the electronic wallet and the location of the consumer against a known-locations database incorporated within the database and comprising data regarding stores, private locations, public places, and transaction data and correlate periodic location patterns to loan and transactional activities and predict by consumer profile and periodicity;
 loan disbursement patterns;   use of loans;   loan repayments; and   transaction activities.