Rolling Window Time-Series Prediction for Real-Time Threshold Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The vast amount of time-series data collected by organizations is difficult to interpret and synthesize, limiting the ability to derive value from it, despite advancements in data collection and storage capabilities.
Innovation Solution
A rolling window approach is used to analyze a limited subset of time-series data, employing statistical predictive modeling techniques like linear regression to generate real-time predictions, with users able to configure actions based on threshold values for predicted metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a rolling window approach is used to analyze a limited subset of time-series data, then real-time prediction generation is enabled, but the ability to analyze the complete historical dataset is reduced
Solution Approach 1:
The patent divides the complete time-series dataset into multiple rolling windows of fixed size. Each window contains a limited subset of recent observations, allowing the predictive model to process manageable portions of data in real-time while maintaining the option to analyze different segments or the complete dataset when needed.
Solution Approach 2:
The rolling window approach dynamically adjusts the data subset being analyzed by continuously moving the window forward in time. This allows the system to adapt to changing data patterns while maintaining a consistent computational workload, enabling real-time predictions without being overwhelmed by the entire historical dataset.
2Measurement precision
If statistical predictive modeling techniques are employed on time-series data, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies statistical predictive modeling techniques to only the relevant portion of data within the rolling window, rather than processing the entire historical dataset. This partial action approach maintains prediction accuracy by focusing on the most recent and relevant observations while significantly reducing computational complexity.
3Extent of automation
If actions are configured based on threshold values for predicted metric values, then proactive responses to metric breaches are enabled, but system complexity increases
Solution Approach 1:
The patent enables users to configure actions based on threshold values for predicted metric values by allowing dynamic parameter specification. Users can define threshold parameters and associated actions without complex system reconfiguration, as the system is designed to accept and process these parameter definitions in a straightforward manner.
Data Source
AI summary
The present approach relates to techniques for efficiently monitoring time-series data for a metric of interest using limited subsets of the time-series data and, based on the modeling, to generate predictions for the metric. Based on the predictions for the metric, one or more actions may be configured to be taken when the predicted value for the metric is outside of a specified range or exceeds a specified threshold.