Prediction Model Training Using Temporal Data Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of predictive analysis decreases when the features of learning data used for training a prediction model differ significantly from those of the prediction data, especially in scenarios with changes in service situations, such as significant changes in service content or emergence of competitors.
Innovation Solution
An information processing method that includes training a prediction model using both learning data and prediction data, where weights are set for learning data samples based on their temporal relationship with prediction data to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If training is performed using only learning data from the past, then the training process is simple, but prediction accuracy deteriorates when service situations change significantly
Solution Approach 1:
The patent applies dynamics by making the training process adaptive rather than static. The learning unit dynamically adjusts the training dataset composition based on the temporal relationship between learning data and prediction data. When service situations change significantly (large temporal gap), the system automatically incorporates prediction data into the training set, transforming the rigid training approach into a flexible, context-aware process that maintains high prediction accuracy under varying conditions.
Solution Approach 2:
The patent changes the parameter of training data composition based on temporal characteristics. By evaluating the temporal relationship between learning data and prediction data, the system adjusts whether to use only learning data or to combine learning data with prediction data for training. This parameter change enables the system to adapt to different service situation stabilities, improving prediction accuracy without requiring complex manual intervention.
2Quantity of substance
If learning data from a long period is used for training, then more training data is available, but feature differences between learning data and prediction data increase
Solution Approach 1:
The patent changes the parameter of data selection from purely quantity-based to quality-and-temporality-based. Instead of always using all available learning data regardless of time span, the system evaluates the temporal relationship and adjusts the training dataset composition accordingly. When the temporal gap is large, prediction data is incorporated to replace or supplement outdated learning data, maintaining feature consistency between training and prediction phases while still utilizing sufficient training samples.
Solution Approach 2:
The patent makes the training dataset composition dynamic rather than fixed. The learning unit continuously evaluates whether the temporal relationship between learning data and prediction data is appropriate, and dynamically adjusts the training approach. This dynamic adjustment ensures that the training data maintains relevant feature characteristics matching the prediction data, preventing feature drift even when using historical data for training.
3Reliability
If prediction data is incorporated into training, then prediction accuracy improves, but additional processing steps are required
Solution Approach 1:
The patent changes the parameter of training data composition based on temporal evaluation rather than always incorporating prediction data. The learning unit first evaluates the temporal relationship between learning data and prediction data, and only incorporates prediction data into training when the temporal gap exceeds a threshold. This conditional approach achieves the benefit of improved prediction accuracy when needed, while avoiding unnecessary processing time when the existing learning data is already sufficient and temporally relevant.
Data Source
AI summary
The present technology relates to an information processing method, an information processing device, and a program capable of improving prediction accuracy of a prediction model.An information processing system including one or more information processing devices performs training of the prediction model on the basis of prediction data used for predictive analysis using the prediction model and learning data. Furthermore, the information processing system including one or more information processing devices performs the predictive analysis on the basis of the prediction model trained on the basis of the learning data and the prediction data, and the prediction data. The present technology can be applied to, for example, a system that performs the predictive analysis for various services.