Imbalanced Data Classification Method Based on Local Mean

Abstract

The invention discloses an unbalanced classification method based on a local mean value. The invention mainly solves the problem that the existing traditional classification algorithm has a low recognition rate for minority classes on an unbalanced data set. The implementation steps are: 1. Input training samples and test samples; 2. Normalization processing; 3. Construct feature weighted vector; 4. Obtain minority class verification samples and majority class verification samples; 5. Obtain verification test sample set and verification Training sample set 6. Calculate and verify the local mean set; 7. Calculate the verification weighted distance; 8. Get the verification result; 9. Determine whether the verification is completed; 10. Get the number of verified local mean values; . Calculate the weighted distance; 13. Obtain the judgment result. The invention effectively improves the recognition rate of minority class samples on the unbalanced data set, and can be applied to the classification and recognition of unbalanced data.

Description

technical field [0001] The invention belongs to the technical field of computer data processing, and further relates to an unbalanced data classification method based on a local mean in the technical field of data classification. The invention can be used in the classification of unbalanced data to improve the recognition rate of minority class data samples. Background technique [0002] Unbalanced data refers to the unbalanced distribution of the number of training samples in different categories. For example, in fault detection, the number of samples with faults is usually small, while the number of samples in normal operation is large. Traditional classification algorithms pursue the overall recognition rate. , naturally tends to the majority class samples, but in practice the minority class samples are the focus of attention, so it is necessary to adjust the traditional classification algorithm to improve the recognition rate of the minority class samples. Imbalanced cl...

AI Technical Summary

Problems solved by technology

The disadvantages of this method are: using active learning to iterate, the process is more complicated, and it is prone to overfitting

The disadvantages of this method are: using fuzzy clustering and self-training of support vector machine to update the sample set, the process complexity is high, and part of the sample information may be lost at the same time

The disadvantage of this method is that when it is applied to unbalanced data classification, the recognition rate will be biased towards the majority of data samples, and the classification performance is unstable when only calculating the local mean of a single sample number.

Images