Multi-Horizon Prediction Model Tuning External Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predicting future values of parameters based on time-series data alone is challenging, especially when considering the impact of external factors, as conventional methods struggle to accurately account for periodic fluctuations and external influences across different time horizons.
Innovation Solution
A multi-horizon predictor system that splits time-series data into training and validation periods for each horizon, tunes a prediction model using machine learning, and adjusts for external data effects to generate specific final models for each horizon, allowing for frequent updates as external data changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional prediction methods use only time series data, then the prediction process is simple, but the prediction accuracy deteriorates when external factors influence the parameter
Solution Approach 1:
The patent combines time series data with external data sources into a unified prediction model. The system integrates multiple data types (historical parameter values and external影响因素) to simultaneously maintain reasonable complexity and improve prediction accuracy, resolving the contradiction between simple processing and accurate results.
Solution Approach 2:
The prediction system is designed to handle both time series data and external data universally. The model can adaptively process different data sources and their interactions, making the system multi-functional in handling various data types while maintaining consistent prediction accuracy across different scenarios.
2Device complexity
If a single prediction model is used for all time horizons, then the model structure is simple, but the prediction accuracy deteriorates across different horizons
Solution Approach 1:
The patent segments the prediction task by time horizon, creating separate prediction models for different horizons (e.g., short-term, medium-term, long-term). Each horizon-specific model can be optimized for its particular time scale, improving overall prediction accuracy while maintaining manageable complexity through modular structure.
Solution Approach 2:
Different prediction models are applied to different time horizons based on local requirements. Each horizon receives a tailored model configuration that optimizes for its specific characteristics, ensuring high prediction accuracy for each local time scale rather than using a one-size-fits-all approach.
3Ease of operation
If external data effects are uniformly adjusted across all horizons, then the tuning process is simple, but the prediction accuracy deteriorates when external data relevance varies by horizon
Solution Approach 1:
The system dynamically adjusts the effect of external data based on the specific time horizon being predicted. The tuning parameters for external data influence vary across horizons, allowing the model to adaptively weight external factors according to their relevance at different time scales, improving accuracy while maintaining operational flexibility.
4Productivity
If the prediction system is updated infrequently, then the computational load is low, but the prediction accuracy deteriorates when external data changes regularly
Solution Approach 1:
The prediction system implements periodic updates at optimized intervals, balancing computational load with the need to incorporate changing external data. By updating predictions periodically rather than continuously or infrequently, the system maintains accuracy in response to external changes while managing computational resources efficiently.
Data Source
AI summary
A multi-horizon predictor system that predicts a future parameter value for multiple horizons based on time-series data of the parameter, external data, and machine-learning. For a given time horizon, a time series data splitter splits the time into training data corresponding to a training time period, and a validation time period corresponding to a validation time period between the training time period and the given horizon. A model tuner tunes the prediction model of the given horizon fitting an initial prediction model to the parameter using the training data thereby using machine learning. The model tuner also tunes the initial prediction model by adjusting an effect of the external data on the prediction to generate a final prediction model for the given horizon using the validation data. A multi-horizon predictor causes the time series data splitter and the model tuner to operate for each of multiple horizons.