Big data-based complex panel data learning method

Abstract

The invention discloses a big data-based complex panel data learning method. While information increment comprised in a new time sequence is fully utilized, data capacity of the time sequence is controlled, so that the calculation cost and the computer storage capacity are controlled; while dimension information comprised in cross section data is fully utilized, support dimension capacity is controlled, so that the calculation cost and the storage capacity are controlled in a computing resource permission range; under the condition of finite panel data capacity, the generalization performanceof learning is ensured, and a generalization theory of an algorithm is given; and complex panel data white noise filtering theory and method are researched to ensure the robustness of a learning model. A dynamic vector fusion technology is adopted; the computing performance is robust; the problem of the high-dimensional complex panel data of the time sequence and the cross section is solved; a linear learning mode is adopted; and online kernel learning for simultaneous control of bidirectional data capacity is realized.

Description

technical field [0001] The invention relates to a panel data learning method, in particular to a panel data learning method based on big data. Background technique [0002] Most of the research on panel data is based on the modeling theory and application of econometrics. Generally, there are assumptions about the data model, and it is a batch data learning model. However, with the complexity of social and economic relations, panel data presents many new characteristics, mainly in the aspects of large data size, intricate relationships, serious data pollution and data loss. In today's big data environment, the original The assumptions of the model are not necessarily true in the context of big data. At the same time, the limitation of computer storage capacity also restricts the normal use of the original panel data model. Contents of the invention [0003] In order to solve the problems of serious pollution and missing data in the existing panel data, the purpose of the...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the problems of serious pollution and missing data in the existing panel data, the purpose of the present invention is to provide a complex panel data learning method based on big data, seek a complex panel data processing mode that reduces computing costs and saves computer storage resources, and uses Solve the problems raised in the above-mentioned background technology