Event prediction using classifier as coarse filter

Abstract

In an aspect, the present application describes a method including: obtaining account data associated with an account; passing at least some of the account data through a non-sufficient funds classifier, the non-sufficient funds classifier being a machine learning classifier configured to classify the account as either likely to have a non-sufficient funds event or unlikely to have a non-sufficient funds event, the machine learning classifier configured to evaluate a plurality of parameters associated with the account data; in response to determining that the account is likely to have a non-sufficient funds event, forecasting a balance for the account, wherein forecasting the balance for the account includes; and providing, to a client device associated with the account, a notification based on a forecasted balance.

What is claimed is: 
     
         1 . A computing system comprising:
 a communications module;   a processor coupled to the communications module; and   a memory storing processor-executable instructions which, when executed, configure the processor to:   pass account data through a non-sufficient funds classifier,   the non-sufficient funds classifier being a machine learning binary classifier that is configured to classify an account associated with the account data as either likely to have a non-sufficient funds event or unlikely to have the non-sufficient funds event, based on an evaluation of a plurality of parameters associated with the account data;   receive a binary output from the machine learning binary classifier indicating that the account is likely to have the non-sufficient funds event;   in response to receiving the binary output indicating that the account is likely to have the non-sufficient funds event, performing an evaluation operation on the account data;   provide, to a client device associated with the account, a notification based on the evaluation operation, the notification including a selectable option to transfer value from another account;   receive a request to transfer value from the another account; and   transfer value from the another account into the account associated with the account data.   
     
     
         2 . The computing system of  claim 1 , wherein the instructions further configure the computing system to train the non-sufficient funds classifier based on a training set by determining respective weights for the plurality of parameters. 
     
     
         3 . The computing system of  claim 2 , wherein the plurality of parameters include one or more of:
 a past non-sufficient funds event; a balance; and an indication of decrease in income.   
     
     
         4 . The computing system of  claim 1 , wherein the notification is provided in response to determining, while performing the evaluation operation, that a forecasted balance is expected to fall below a defined threshold. 
     
     
         5 . The computing system of  claim 4 , wherein the defined threshold is zero. 
     
     
         6 . The computing system of  claim 1 , wherein performing an evaluation operation on the account data includes determining one or more projected future transfers, and wherein the one or more projected future transfers include at least one incoming transfer and at least one outgoing transfer. 
     
     
         7 . The computing system of  claim 1 , wherein the account data passed through the machine learning binary classifier includes a transaction listing. 
     
     
         8 . The computing system of  claim 1 , wherein the account data passed through the machine learning binary classifier includes biographical data. 
     
     
         9 . The computing system of  claim 1 , wherein the evaluation operation performed in response to receiving the binary output indicating that the account is likely to have the non-sufficient funds event is not performed when a binary output is received indicating that the account is not likely to have the non-sufficient funds event. 
     
     
         10 . A method of notifying of a predicted condition, the method comprising:
 passing account data through a non-sufficient funds classifier, the non-sufficient funds classifier being a machine learning binary classifier that is configured to classify an account associated with the account data as either likely to have a non-sufficient funds event or unlikely to have the non-sufficient funds event, based on an evaluation of a plurality of parameters associated with the account data;   receiving a binary output from the machine learning binary classifier, the binary output indicating that the account is likely to have the non-sufficient funds event;   in response to receiving the binary output indicating that the account is likely to have the non-sufficient funds event, performing an evaluation operation on the account data;   providing, to a client device associated with the account, a notification based on the evaluation operation, the notification including a selectable option to transfer value from another account;   receiving a request to transfer value from the another account; and   transferring value from the another account into the account associated with the account data.   
     
     
         11 . The method of  claim 10 , further comprising training the non-sufficient funds classifier based on a training set by determining respective weights for the plurality of parameters. 
     
     
         12 . The method of  claim 10 , wherein the plurality of parameters include one or more of:
 a past non-sufficient funds event;   a balance; and   an indication of decrease in income.   
     
     
         13 . The method of  claim 10 , wherein the notification is provided in response to determining, while performing the evaluation operation, that the forecasted balance is expected to fall below a defined threshold. 
     
     
         14 . The method of  claim 13 , wherein the defined threshold is zero. 
     
     
         15 . The method of  claim 10 , wherein performing an evaluation operation on the account data includes determining one or more projected future transfers, and wherein the one or more projected future transfers include at least one incoming transfer and at least one outgoing transfer. 
     
     
         16 . The method of  claim 10 , wherein the account data passed to through the machine learning binary classifier includes a transaction listing. 
     
     
         17 . The method of  claim 10 , wherein the account data passed through the machine learning binary classifier includes biographical data. 
     
     
         18 . The method of  claim 10 , wherein the evaluation operation performed in response to receiving the binary output indicating that the account is likely to have the non-sufficient funds event is not performed when a binary output is received indicating that the account is not likely to have the non-sufficient funds event. 
     
     
         19 . A non-transitory computer-readable storage medium comprising processor-executable instructions which, when executed, configure a processor to:
 pass second account data through a non-sufficient funds classifier, the non-sufficient funds classifier being a machine learning binary classifier that is configured to classify the account as either likely to have a non-sufficient funds event or unlikely to have the non-sufficient funds event, based on an evaluation of a plurality of parameters associated with the account data;   receive a binary output from the machine learning binary classifier indicating that the account is likely to have the non-sufficient funds event;   in response to receiving the binary output indicating that the account is likely to have the non-sufficient funds event, performing an evaluation operation on the account data;   provide, to a client device associated with the account, a notification based on the evaluation operation, the notification including a selectable option to transfer value from another account;   receive a request to transfer value from the another account; and   transfer value from the another account into the account associated with the account data.   
     
     
         20 . The non-transitory computer-readable storage medium of claim of  claim 19 , wherein the evaluation operation performed in response to receiving the binary output indicating that the account is likely to have the non-sufficient funds event is not performed when a binary output is received indicating that the account is not likely to have the non-sufficient funds event.