Event Prediction Model Input Selection Using Variable Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing event prediction models rely heavily on subject matter expert input, which limits the selection of input data variables and often fails to include the most predictive variables, leading to inaccurate and untimely predictions due to incomplete knowledge of all potential causes and environmental changes.
Innovation Solution
A data-driven approach is employed by the asset data platform to select input data variables for event prediction models, involving the classification and scoring of initial data variables based on historical data, filtering, and applying transformations to enhance predictive accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If subject matter expert input is used to select input data variables, then the model development process is simplified and faster, but the selection of input data variables is limited and may miss the most predictive variables
Solution Approach 1:
The system automatically selects input data variables using algorithmic processes that evaluate historical data and predictive power without requiring manual subject matter expert intervention. The computer system performs variable selection, scoring, and transformation automatically, allowing the system to serve itself in the variable selection process while improving prediction accuracy through data-driven methods.
2Reliability
If a comprehensive set of all possible input data variables is included in the model, then the model may capture all potential causes, but the computational cost increases and model complexity increases
Solution Approach 1:
The system extracts only the most relevant and predictive variables from the comprehensive set of all possible input data variables. Through automated scoring and filtering processes, the system identifies and extracts the subset of variables that contribute most to prediction accuracy, eliminating redundant and less predictive variables to reduce model complexity while maintaining prediction completeness.
Solution Approach 2:
The system transforms and modifies input variables through various transformations to enhance their predictive power. By changing the form and parameters of the data variables through automated transformations, the system improves model performance with a optimized set of variables rather than requiring all possible variables in their original form.
3Reliability
If irrelevant or marginally-relevant variables are included in the model, then the model may be more comprehensive, but computational cost increases and model performance degrades
Solution Approach 1:
The system applies a filtering process that goes beyond simple inclusion of variables to selectively retain only those variables that meet predetermined thresholds for predictive relevance. By applying excessive filtering criteria and automatically eliminating variables that do not meet the threshold, the system achieves computational efficiency while maintaining prediction robustness through data-driven variable selection.
Data Source
AI summary
Disclosed is a process for creating an event prediction model that employs a data-driven approach for selecting the model's input data variables, which, in one embodiment, involves selecting initial data variables, obtaining a respective set of historical data values for each respective initial data variable, determining a respective difference metric that indicates the extent to which each initial data variable tends to be predictive of an event occurrence, filtering the initial data variables, applying one or more transformations to at least two initial data variables, obtaining a respective set of historical data values for each respective transformed data variable, determining a respective difference metric that indicates the extent to which each transformed data variable tends to be predictive of an event occurrence, filtering the transformed data variables, and using the filtered, transformed data variables as a basis for selecting the input variables of the event prediction model.