Prediction Model Update via Error Threshold Notification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction models for time series data often require frequent updates due to their inability to accurately reflect future changes, leading to high computation loads and challenges in selecting an appropriate historical data range for updating, without clear notification of when updates are needed.
Innovation Solution
A method and apparatus for updating prediction models by selecting a prediction time window, calculating prediction errors, and providing notifications when errors exceed a predefined threshold, while also determining an optimal historical data range for updating based on accumulative errors and predefined ratios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the prediction model is updated frequently to maintain accuracy, then the prediction accuracy is improved, but the computation load increases significantly
Solution Approach 1:
The patent implements a feedback mechanism by calculating prediction errors (MAPE) between predicted values and actual measured values, then using this error information to determine whether model updating is necessary. This feedback loop enables the system to maintain prediction accuracy while avoiding unnecessary frequent updates that would increase computation load.
Solution Approach 2:
The patent performs preliminary error calculation and threshold comparison before actually updating the prediction model. By evaluating prediction errors against predefined thresholds in advance, the system determines whether model updating is needed, thereby avoiding unnecessary computation associated with frequent model updates.
2Measurement precision
If the prediction model is updated frequently to maintain accuracy, then the prediction accuracy is improved, but the device complexity increases
Solution Approach 1:
The system uses prediction error feedback to control model updating decisions. By continuously monitoring whether prediction errors exceed thresholds, the system simplifies the update process to only occur when necessary, reducing the overall complexity of the model update mechanism while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary evaluation of prediction errors before triggering model updates. This preliminary action simplifies the device complexity by adding a decision layer that prevents unnecessary updates, making the overall system easier to manage despite the capability for frequent updates when needed.
3Measurement precision
If a large range of historical data is used for model updating, then the prediction accuracy is improved, but the time required for data processing increases
Solution Approach 1:
The patent dynamically adjusts the training time window parameters based on prediction errors. When errors exceed thresholds, the system modifies the historical data range parameters for model retraining, optimizing the balance between using sufficient historical data for accuracy and limiting the data range to reduce processing time.
Solution Approach 2:
The system performs preliminary calculation of accumulative errors over different time windows before selecting the optimal training data range. This preliminary error analysis enables the system to identify the most relevant historical data period, reducing unnecessary data processing while maintaining prediction accuracy.
4Measurement precision
If the training time window is extended to include more historical data, then the prediction accuracy is improved, but the computation load increases
Solution Approach 1:
The patent dynamically adjusts the training time window parameter based on prediction error thresholds. When errors indicate model degradation, the system modifies the training data range parameters to include appropriate historical data, optimizing the balance between accuracy improvement and computation load management.
Solution Approach 2:
The system performs preliminary error calculation and threshold comparison before extending the training time window. By evaluating whether prediction errors warrant model updates in advance, the system avoids unnecessary computation associated with extending training data ranges, thereby reducing overall computation load while maintaining accuracy when needed.
Data Source
AI summary
Updating a prediction model, where the prediction model is used for time series data, a computer selects a first prediction time window in an order from a plurality of prediction time windows associated with the prediction model, and predicts predicted values of the time series data at time points within the first prediction time window. The computer calculates a prediction error associated with the first prediction time window based on the one or more predicted values and one or more actual measured values of the time series data at the plurality of time points. The computer determines whether the prediction error is larger than a predefined error threshold associated with the first prediction time window, and in response to determining the prediction error is larger than the predefined error threshold, provides a notification of updating the prediction model.