Training data resampling method, device, storage medium and electronic equipment

Abstract

Disclosed are a training data re-sampling method and apparatus, and a storage medium and an electronic device. The method comprises: acquiring first original data within a first time period (S101); calculating respective first proportions of multiple pre-set classifications in the first original data (S102); sorting the multiple pre-set classifications according to a size relationship of the first proportions and a pre-set rule so as to obtain a first sorting result (S103); determining, according to the first sorting result of the pre-set classifications and a pre-set correlation, a sampling proportion corresponding to each pre-set classification (S104), wherein the pre-set correlation is a correlation between the first sorting result and the sampling proportion; and re-sampling training data for modeling according to the sampling proportions respectively corresponding to the multiple pre-set classifications (S105). The problem of a classification model being unfriendly to a small category can be solved, and the classification accuracy of a classification model, obtained through training by means of training data, for different applications can be improved, thereby improving user experience.

Description

technical field [0001] The present disclosure relates to the field of data mining, in particular, to a training data resampling method, device, storage medium and electronic equipment. Background technique [0002] In machine learning, the number of samples in the training data for different classifications in the classification model may often vary greatly. For example, in N training data, the number of samples belonging to the first class may be different from the number of samples belonging to the second class. The number of samples and the number of samples belonging to the third category are very different (for example, the number of samples belonging to the first category may account for 90% of the N training data, and the number of samples belonging to the second and third categories may only account for 90% of the N training data. 10% of the N training data), so that when the training data with an unbalanced number of samples is directly used to train the classificat...

AI Technical Summary

Problems solved by technology

[0003] The purpose of this disclosure is to provide a training data resampling method, device, storage medium and electronic equipment, which can resample the training data according to the proportion of different classifications in the actual original data in view of the unbalanced training data processing, so as to solve the problem that the classification model is not friendly to small categories

Images