Multivariate AI Modeling for Holistic Negative Cash-Flow Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional computer-implemented methods for predicting customer/account cash flow are univariate and unable to leverage all available information, requiring high computational capacity and processing time due to analyzing data at an account level, thus lacking holistic analysis.
Innovation Solution
A customized multivariate deep neural network is used to analyze user account data, predicting negative cash flow, depth, and duration, trained on historical data from a subset of users, and integrated with an analytical model to determine account needs and establish electronic communication sessions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional univariate time series forecasting models are used to analyze account data, then the analysis focuses on single account data, but the accuracy is insufficient and computational resources are wasted due to inability to leverage all available information
Solution Approach 1:
The patent combines multiple univariate time series models into a single multivariate deep neural network that processes multiple accounts simultaneously. This merging allows the system to leverage correlations between accounts and utilize all available information across the user's account portfolio, thereby improving prediction accuracy while reducing redundant computation.
Solution Approach 2:
The multivariate deep neural network is designed to handle multiple accounts and multiple predictive tasks simultaneously. Instead of creating separate models for each account, the system uses a universal model that can analyze any account within a user's portfolio, making the system more efficient and accurate by learning from diverse data patterns across all accounts.
2Reliability
If conventional software solutions analyze data at account level individually, then each account receives focused analysis, but high computational capacity and processing time are required
Solution Approach 1:
The patent merges the analysis of multiple accounts into a single multivariate deep neural network processing step. By combining accounts that belong to the same user into one analysis unit, the system achieves holistic analysis completeness while dramatically improving processing efficiency through shared computational resources and parallel processing capabilities.
Solution Approach 2:
The system segments users into groups based on their account portfolios and applies the multivariate deep neural network at the user level rather than individual account level. This segmentation strategy allows the system to maintain comprehensive analysis coverage while reducing the total number of model executions needed, thereby improving overall processing efficiency.
3Adaptability or versatility
If conventional time series models are used, then the system is limited to single account analysis, but holistic analysis of user data cannot be provided
Solution Approach 1:
The multivariate deep neural network is designed as a universal model that can analyze multiple accounts simultaneously and provide holistic user-level insights. This single model structure replaces the need for multiple separate univariate models, expanding the analysis scope to cover entire user portfolios while managing complexity through a unified architectural approach.
Solution Approach 2:
The patent transitions from univariate (single variable per account) to multivariate (multiple variables across multiple accounts) analysis by adding the dimension of account interrelationships. This dimensional expansion allows the system to capture correlations and patterns across accounts, providing holistic analysis while the structured neural network architecture manages the increased complexity through systematic feature engineering and processing.
Data Source
AI summary
Disclosed methods and system describe a server that uses AI modeling to predict negative cash flow at a user level. The server periodically retrieves data associated with the user, the data comprising monetary attributes associated with one or more accounts of the user; executes a deep neural network model trained based upon historical data associated with at least a subset of the users configured to predict a negative cash flow in one or more accounts of the user, a depth of the negative cash flow, and a duration of the negative cash flow; transmits, to a second server, the predicted values, whereby when the second server determines that a likelihood of account needs satisfies a threshold, the second server establishes an electronic communication session with an electronic device of the user; trains the deep neural network when the second server establishes the electronic communication session.