Deep learning-based missing value filling method and system

Abstract

The invention provides a deep learning-based missing value filling method and system. The method comprises the following steps of preprocessing a data set; training and storing a preliminarily constructed convolutional neural network by utilizing a training sample set, carrying out missing value filling on a missing test sample set by using the trained convolutional neural network, comparing the filling result with a test sample set, and adjusting a network structure of the convolutional neural network and iterating the training and verification steps until a precision requirement met when the precision requirement is not met; inputting a complete data subset into the convolutional neural network so as to obtain a perfect convolutional neural network; and inputting a missing data subset into the perfect convolutional neural network to complete missing value filling. According to the method and system provided by the invention, the database missing value filling problem is solved, the effects of higher correctness and higher efficiency are achieved, and the missing data can be restored more really and rapidly.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method and system for filling missing values ​​based on deep learning. Background technique [0002] Since information technology has been widely applied to various industries and promoted the development of these new and old fields at a super speed, data, as the resource on which this technology depends, has been continuously collected and mined, and the amount of data is expanding at an alarming rate. Huge data undoubtedly increases the difficulty of data management. In the real world, due to omissions in data entry, incorrect measurement methods, restrictions on collection conditions, or deletion due to violation of constraints, many factors may lead to missing data. Missing values ​​not only mean blank information, but more importantly, it will affect subsequent data mining, statistical analysis and other work. Common ways to deal with missing values ​​include r...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to provide a missing value filling method based on deep learning in view of the fact that the missing value filling method in the prior art relies heavily on the integrity of the existing data and cannot find the deep relationship between the data. The value filling method and system, using the deep neural network, can deeply mine the characteristics of the internal and interrelationships of the data, and can improve the filling accuracy and filling efficiency at the same time

Images