Time Series Prediction Using Media Event Impact
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Purely mathematical models for predicting future system behavior from time series data fail to adequately handle unusual events unrelated to the system being measured, as they do not account for the business impact of real-world events.
Innovation Solution
The system gathers information about real-world events from external sources like social network sites and news aggregator sites, classifies this information, and uses it to measure the effect of these events on metrics, incorporating this impact into predictions using a combination of network content crawling, classification, learning, and prediction services.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If purely mathematical models are used for prediction, then the prediction system is simple and easy to implement, but the prediction accuracy deteriorates when unusual real-world events occur
Solution Approach 1:
The patent introduces media events as an intermediary element that bridges the gap between mathematical models and real-world phenomena. The system collects, classifies, and integrates media event data from external sources as a mediator that adjusts prediction outcomes based on real-world events, thereby improving prediction reliability without requiring complete system redesign
Solution Approach 2:
The patent merges two previously separate systems: the mathematical time-series prediction model and the media event analysis system. By combining these systems into a unified prediction framework where media events modulate mathematical model outputs, the patent achieves both improved accuracy and controlled complexity through integration
2Measurement precision
If media event data is collected from external sources, then the prediction accuracy improves, but the data collection and processing complexity increases
Solution Approach 1:
The patent segments the data processing workflow into distinct modular components: data collection from external sources, classification of media events by category and sentiment, impact measurement on specific metrics, and integration with prediction models. This segmentation reduces processing complexity by making each component independent and manageable
Solution Approach 2:
The patent performs preliminary classification and categorization of media events before they are used in prediction. By pre-processing event data into structured formats with assigned categories and sentiment scores in advance, the system reduces the complexity of real-time processing during prediction operations
3Reliability
If real-world events are incorporated into the prediction model, then the forecast reliability improves, but the computational resources required increase
Solution Approach 1:
The patent applies partial action by selectively incorporating only relevant media events into the prediction model based on their classification and measured impact. Rather than processing all available event data, the system focuses on events that have significant impact on specific metrics, reducing computational energy requirements while maintaining forecast reliability
Data Source
AI summary
Disclosed are various embodiments for using media reported events to generate predictions for time series data. Information retrieved from a plurality of network content sources is classified into a plurality of categories. A prediction is generated for a time series. The time series is associated with a metric observed in a computing system. The generated prediction takes into account an impact of at least one of instance of the classified information.