| CPC G06Q 30/0255 (2013.01) [G06Q 30/0201 (2013.01); G06Q 30/0277 (2013.01)] | 20 Claims |
|
1. A computing system, comprising:
a memory;
one or more processors in communication with the memory; and
program instructions executable by the one or more processors via the memory to:
perform predictive analytics using one or more analytical tools on an aggregation of user action history data that incorporates financial transactions previously performed by a user, the financial transactions including loan payments made by the user, the predictive analytics comprising:
classifying, using a cluster model, the user action history data including the financial transactions previously performed by the user, the classifying comprising assigning a probability score indicating a likelihood one or more of the financial transactions are to be assigned to one or more clusters of data;
predicting, based on the classifying and the user action history data of the financial transactions, a likely interest rate being applied to an existing liability product the user is making payments towards from the loan payments made by the user as indicated by the aggregation of user action history data;
performing a credit evaluation that includes automatically generating, based on the classifying and the user action history data, a predicted credit score that is predicted to have applied when the user initiated a transaction incorporating the existing liability product; train, using a set of training data of a plurality of users, a neural network,
the neural network utilizing interconnected nodes, to make predictions to identify a likely future outcome resulting from displaying interactive elements, where the likelihood of the likely future outcome is informed by data attribute classifications attributed to the plurality of users, where the interactive elements suggest entity products, the training incorporating an iterative training and testing loop that includes:
repeatedly, in each training iteration, simulating predicted input responses by users that are predicted to result from displaying, via a digital platform, the interactive elements;
testing and comparing, in each training iteration, the predicted input responses against a target variable to be predicted;
indicating, via a feedback loop, in each training iteration, weights applied to nodes of the interconnected nodes to be modified to improve predictability of the target variable and reduce an error amount of output data compared to the target variable;
updating calculations used to predict the target variable by adjusting the weights applied to the nodes, the weights indicating an impact a node in a layer of the neural network has on a connected node in another layer of the neural network, thereby reducing the error amount and improving predictability of the target variable accuracy of predictions made by the neural network and improving function of the computing system;
deploy, based on the error amount being within an acceptable level, the trained neural network;
apply the deployed neural network to user data of the user to predict the interactive elements that suggest the entity products that would be most likely to align with the likely future outcome the neural network is trained to predict, the user data including the likely interest rate predicted and the predicted credit score, where the predicted input responses leading to the likely future outcome are capable of being at least partially performed across a network via the digital platform, and wherein the likely future outcome comprises replacing the existing liability product the user is making payments toward with one of the entity products, where the entity products are provided by an entity associated with the digital platform;
assign an engagement probability score to each of the predicted interactive elements that suggest the entity products, the engagement probability score representing a likelihood of a predicted input response;
determine that the user is accessing, via a user interface of a user device, the digital platform to perform one or more user actions; and
display, via the user interface, the interactive elements, the interactive elements being displayed as a list included in an aggregation table of recommended entity products, the interactive elements being prioritized within the list in accordance with the assigned engagement probability score such that the prioritized interactive elements are most likely to align with the likely future outcome the neural network is trained to predict, the displaying of the interactive elements to facilitate performance of the predicted input response that is capable of being at least partially performed, via the digital platform, where the performance of the predicted input response initiates the likely future outcome of replacing the existing liability product the user is making payments toward with an entity product of the entity products.
|