Automatic selection system and method for time series data prediction algorithms

Abstract

The invention belongs to the technical field of time series data prediction algorithms, in particular to an automatic selection system for time series data prediction algorithms. The automatic selection system comprises a control chip. The invention further discloses an automatic selection method for the time series data prediction algorithms. The automatic selection method for the time series data prediction algorithms comprises the following steps of: S1, converting a problem space into a sub-time series space instance; S2, converting the sub-time series space instance into a feature set instance through utilizing a feature selection sub-protocol; S3, realizing mapping with an algorithm instance and a computing performance instance; S4, and establishing a corresponding relationship between a problem instance and the calculation performance instance. When processing a prediction time series, a relationship among an instance data set space, a feature set space, an algorithm space and acalculation performance index space is established, thus the prediction algorithm input into the instance data set is recommended according to the calculation performance index space.

Description

technical field [0001] The invention relates to the technical field of time series data prediction algorithms, in particular to a system and method for automatically selecting time series data prediction algorithms. Background technique [0002] There are many types of existing time series data prediction algorithms. In the case of limited computing performance, it is not possible to use all algorithms to calculate at the same time, and then compare the results for data prediction. Therefore, it is necessary to artificially control the type of algorithm used. However, in the process of processing a large amount of data, it takes a lot of time. Using an intelligent database for data processing can effectively improve the overall data budget speed and the accuracy of prediction results. Therefore, the present invention aims to solve multi-type time series, Mutual matching of multi-class prediction algorithms and multi-class sequence feature sets under the condition of limited ...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a system and method for automatically selecting time series data forecasting algorithms, to solve the problem of how to solve the problem of multi-class time series, multi-class forecasting algorithms, and multi-class sequence feature sets in the above-mentioned background technology. The problem of mutual matching in the case of limited computing performance

Images