Explainable ML Time-Series Transformation for Risk Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models fail to accurately predict outcomes by only evaluating input variables at a single time point, neglecting the temporal changes in these variables.
Innovation Solution
The development of an explainable machine learning model that utilizes transformed time-series data, applying multiple families of transformations to generate multiple sets of transformed data instances, which are then input into a neural network model to determine risk indicators with a monotonic relationship.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If static models evaluate input variables at a single time point, then the model complexity is low, but the prediction accuracy is insufficient
Solution Approach 1:
The patent segments the time-series data into multiple time points and applies different transformation families (e.g., polynomial transformations of different degrees, transformations with different time windows) to create multiple sets of transformed data instances. This segmentation allows the model to capture temporal patterns at different granularities without requiring a single overly complex model structure.
Solution Approach 2:
The patent transforms the original time-series data into multiple dimensions by applying different transformation families, creating transformed data instances that represent the same underlying variable from different temporal perspectives. This dimensional expansion enables the model to capture complex temporal relationships while maintaining interpretability through the monotonic relationship constraint.
2Measurement precision
If multiple families of transformations are applied to time-series data, then the prediction accuracy is enhanced, but the data processing complexity increases
Solution Approach 1:
The patent changes parameters of the transformation families (e.g., polynomial degree, time window size, transformation type) to generate diverse transformed data instances from the same input time-series data. By systematically varying transformation parameters, the model captures multiple aspects of temporal behavior without manually creating complex processing pipelines for each scenario.
Data Source
AI summary
Various aspects involve explainable machine learning based on time-series transformation. For instance, a computing system accesses time-series data of a predictor variable associated with a target entity. The computing system generates a first set of transformed time-series data instances by applying a first family of transformations on the time-series data. Any non-negative linear combination of the first family of transformations forms an interpretable transformation of the time-series data. The computing system determines a risk indicator for the target entity indicating a level of risk associated with the target entity by inputting the first set of transformed time-series data instances into a machine learning model. The computing system transmits, to a remote computing device, a responsive message including the risk indicator. The risk indicator is usable for controlling access to one or more interactive computing environments by the target entity.