Predictive Model Decay Rate Adjustment for Forecast Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing predictive models struggle to accurately determine the forecasted performance level of option data objects, such as deal offers, due to their reliance on outdated time-series data that does not reflect current operating conditions, especially when changes in operating parameters occur frequently, leading to inaccurate predictions and increased computational and storage demands.
Innovation Solution
The system calculates an adjusted decay rate for time-series data associated with option data objects prior to an event, allowing for the effective weighting of current data and reducing the influence of outdated information by applying a second decay rate to time-series data related to the event, thereby reflecting the change in operating conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional predictive models use all historical time-series data with uniform weighting, then the model maintains simplicity in calculation, but the prediction accuracy deteriorates due to outdated data not reflecting current operating conditions
Solution Approach 1:
The patent applies different decay rates to different segments of historical data based on operating condition changes. When an event indicating a parameter change is detected, the system adjusts the decay rate applied to data before versus after that event, allowing recent data to have higher weight than outdated data without requiring manual intervention or complex algorithms
Solution Approach 2:
The patent segments the historical time-series data into multiple periods based on detected events. Each segment is assigned a different decay rate, creating a piecewise weighting scheme that simplifies the handling of varying data relevance while improving prediction accuracy by distinguishing between outdated and current operating conditions
2Reliability
If the system applies a single decay rate to all historical data, then the computational process remains simple and fast, but the forecasted performance level becomes inaccurate when operating parameters change
Solution Approach 1:
The system pre-calculates and stores event indicators that mark changes in operating parameters. When making predictions, it only needs to retrieve these pre-identified event points and apply corresponding decay rates, avoiding the need for real-time analysis of whether parameters have changed and maintaining processing efficiency
Solution Approach 2:
The decay rate parameter is dynamically adjusted based on detected events. Before an event, one decay rate is applied; after an event, a different decay rate is applied. This automatic parameter adjustment improves forecast reliability without significantly increasing computational burden
3Reliability
If outdated time-series data is given equal weight as current data, then data storage requirements remain low and processing is simplified, but prediction reliability deteriorates due to influence from obsolete operating conditions
Solution Approach 1:
The patent changes the weighting parameter (decay rate) applied to historical data based on whether it predates or postdates an operating condition change event. This creates an automatic differentiation between relevant and obsolete data without requiring physical data filtering or selective storage
Solution Approach 2:
The patent extracts and applies different decay rates to different portions of the historical data based on event boundaries. By separating the data weighting into distinct segments, it effectively removes the harmful influence of outdated data while retaining useful information from all time periods
Data Source
AI summary
An apparatus, method, and computer program product are provided to adjust and modify input signals used in connection with predictive models by detecting events, such as changes in operating parameters of data objects and/or related systems and calculating adjusted decay rates to be applied to time-series data associated with times prior to an occurrence of an event. In some example implementations, an indication of an event associated with a given datastream is received, in a manner which indicates the change in an operating parameter and the time at which the change occurred. Based at least in part on the indication of the event associated with the datastream, a second decay rate associated with the set of time-series data is determined and applied to the set of time-series data, such that an updated future performance level can be calculated by a predictive model.