Ensemble Time Series Forecasting with Multi-Source Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning predictive models often perform sub-optimally due to limited availability of training data, leading to inefficiencies in making predictions based on time series data.
Innovation Solution
An ensemble time series prediction system that leverages multiple categories of data, including time-dependent, embedding, and time-lagged features, using sub-models like LSTM for sequential data processing, and consolidates these features for input into a recurrent neural network to generate predictions with confidence intervals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If a machine learning predictive model uses limited historical time series data for training, then the model can be trained with available data, but the prediction accuracy deteriorates due to insufficient training data
Solution Approach 1:
The patent combines multiple heterogeneous data sources (time series data, tabular data, unstructured data) into a unified training dataset. This merging of diverse data types increases the effective quantity of training data available to the machine learning model, thereby improving prediction accuracy without requiring more historical time series data alone
Solution Approach 2:
The system creates a multi-functional data processing framework that handles various data types (structured, unstructured, time-series, tabular) through a single unified model architecture. This universal approach allows the model to learn from diverse data sources simultaneously, compensating for limited domain-specific training data
2Measurement precision
If an ensemble model with multiple sub-models is constructed to process different data categories, then the prediction performance improves, but the system complexity increases
Solution Approach 1:
The patent segments the data processing into specialized sub-models that each handle specific data types (e.g., one sub-model for time series data, another for tabular data). This segmentation allows each sub-model to be optimized for its specific data category while maintaining overall system manageability through modular architecture
Solution Approach 2:
The system introduces intermediate feature extraction layers that transform heterogeneous data from different sources into a unified feature representation. These intermediary layers act as mediators between diverse input data and the final prediction model, simplifying the integration process and reducing overall system complexity
Data Source
AI summary
An ensemble time series prediction system that makes predictions based on observed data. The disclosed ensemble time series prediction system may leverage different types of datasets and information from different resources for making predictions. The disclosed ensemble time series prediction system may extract time dependent features from autoregressive time dependent data, embedding features from sparse datasets, continuous features from continuous dataset, and time lagged features from data that include time-lag information. The disclosed ensemble time series prediction system may then consolidate the features extracted from the different types of datasets and generate a set of consolidated input features for training a neural network, which may include a recurrent neural unit that finds sequential pattern for the sequence of input features and a regression unit that performs regression and predictions. The ensemble time series prediction system may output a set of outputs that include predicted values and associated confidence intervals.