Multivariate Time-Series Forecasting to Reduce Spurious Correlations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional time-series forecasting methods, such as ARIMA and ETS models, are limited by their reliance on linear relationships, univariate data, and one-step predictions, and can be inaccurate due to spurious correlations when making multi-step forecasts.
Innovation Solution
A monitoring device uses multiple machine learning models (e.g., CNN, RNN, LSTM) to generate and combine multi-step forecast values, employing voting and averaging techniques to minimize spurious correlations and enhance accuracy, while also reducing resource usage by identifying resource correlations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional linear models (ARIMA, ETS) are used for time-series forecasting, then the models are lightweight and require less data, but they produce inaccurate multi-step forecasts due to spurious correlations and inability to capture complex patterns
Solution Approach 1:
The patent combines multiple machine learning models (CNN, RNN, LSTM) into an ensemble forecasting system. Each model processes the same time-series input data independently, and their predictions are aggregated through voting or averaging mechanisms. This merging of multiple models captures diverse patterns in the data while maintaining robustness against individual model failures, thereby improving forecast accuracy for multi-step predictions without requiring any single model to be overly complex.
2Measurement precision
If multiple machine learning models are used to generate forecasts, then forecast accuracy improves, but computational resources and processing time increase
Solution Approach 1:
The patent extracts and eliminates redundant computational efforts by implementing a voting mechanism that identifies and discards outlier predictions from the ensemble models. Instead of processing and storing all predictions from multiple models, the system extracts only the consensus forecast values that agree across models, removing spurious correlations and unnecessary computational overhead. This extraction approach maintains high accuracy while reducing the energy and computational resources required for forecast generation.
3Reliability
If multiple time-series data streams from multiple resources are processed, then comprehensive forecasting is achieved, but data processing overhead and resource consumption increase
Solution Approach 1:
The patent performs preliminary correlation analysis on time-series data streams before feeding them into the machine learning models. By pre-processing the data to identify and remove highly correlated or redundant data streams, the system reduces the volume of input data without compromising forecast reliability. This preliminary action eliminates unnecessary data processing overhead while maintaining the comprehensive nature of the forecasting approach, thereby improving processing efficiency.
Data Source
AI summary
In some implementations, a monitoring device may obtain a plurality of time-series data streams respectively associated with a plurality of resources. The monitoring device may generate, using a plurality of machine learning models and based on the plurality of time-series data streams, a plurality of sets of multi-step forecast values, wherein each set of multi-step forecast values is associated with the plurality of resources. The monitoring device may determine, based on the plurality of sets of multi-step forecast values, a set of particular multi-step forecast values associated with the plurality of resources. The monitoring device may cause, based on the set of particular multi-step forecast values, one or more actions to be performed. In some implementations, the monitoring device may determine, based on the plurality of time-series data streams and the plurality of sets of multi-step forecast values, that a correlation exists between a first resource and a second resource.