Time Series Forecasting Mechanism Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for time series data forecasting in computing and communications systems face challenges in selecting the optimal forecasting mechanism, leading to inaccurate forecasts and potential performance issues due to the use of non-optimal techniques.
Innovation Solution
A system and method that receive time series data, determine if the data point count meets a threshold, generate test forecasts using various forecasting mechanisms, compare these forecasts to a subset of data points to evaluate errors, and select the optimal mechanism based on these errors for generating an output forecast.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple forecasting mechanisms are tested and evaluated, then the accuracy of the forecast is improved, but the computational time and complexity increase
Solution Approach 1:
The system performs preliminary testing of multiple forecasting mechanisms using a training subset of data before generating the final forecast. This preliminary action allows the system to evaluate and select the most accurate mechanism in advance, ensuring high forecast accuracy while managing computational time through efficient pre-computation.
Solution Approach 2:
The system implements feedback by comparing the forecasts generated by different mechanisms against actual data points to determine errors. This feedback loop enables the system to select the mechanism with the lowest error, continuously improving forecast accuracy while optimizing computational resources through iterative evaluation.
2Reliability
If multiple forecasting mechanisms are evaluated using test forecasts, then the reliability of the selected mechanism is improved, but the device complexity increases
Solution Approach 1:
The system segments the data into training and testing subsets, and separately evaluates different forecasting mechanisms on these subsets. This segmentation allows for systematic comparison of mechanisms without requiring complex integration of all possible forecasting approaches, thereby improving reliability while managing complexity through structured evaluation.
Solution Approach 2:
The system changes parameters such as the number of data points used for training and testing, and the criteria for selecting the optimal mechanism. By adjusting these parameters, the system can optimize the balance between evaluation thoroughness (reliability) and computational complexity, ensuring robust mechanism selection without excessive system complexity.
3Measurement precision
If a threshold for data point count is enforced, then the quality of the forecast is improved, but the availability of data is reduced
Solution Approach 1:
The system applies different quality thresholds and evaluation criteria to different segments of the data. By ensuring that only data meeting specific quality thresholds are used for training and testing, the system maintains high forecast quality while maximizing the utilization of available data, rather than discarding potentially useful data points.
Data Source
AI summary
Time series data is generated and forecasted with a selected forecasting mechanism. Time series data to forecast including a plurality of data points is received. A count of the plurality of data points is determined to meet a threshold. Responsive to that determination, a plurality of test forecasts are generated with respective forecasting mechanisms of a plurality of forecasting mechanisms using a first subset of the plurality of data points. Errors are then determined for the respective forecasting mechanisms, such as based on comparisons of corresponding ones of the plurality of test forecasts and a second subset of the plurality of data points. One of the plurality of forecasting mechanisms is selected based on the errors. An output forecast is then generated with the selected forecasting mechanism using the first and second subsets of the plurality of data points.