Improved SMOTE re-sampling method for unbalanced data classification

A technology of unbalanced data and data, which is applied in the direction of instruments, character and pattern recognition, computer components, etc., can solve the problems of blindness and marginalization of neighbor selection, reduce overlap between sampling classes, reduce sampling interference, and reduce interference Effect

Inactive Publication Date: 2017-11-07
NANJING UNIV OF POSTS & TELECOMM
View PDF0 Cites 15 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problem of blindness and marginalization of neighbor selection in the traditional SMOTE method, and an improved K-Means-SMOTE method is proposed for this purpose

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Improved SMOTE re-sampling method for unbalanced data classification
  • Improved SMOTE re-sampling method for unbalanced data classification
  • Improved SMOTE re-sampling method for unbalanced data classification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0026] In order to better illustrate the unbalanced data set re-sampling method involved in the present invention, it is applied to the complaint model of Internet TV set-top box users in the following. In this type of model, the data is divided into two categories: the first category is set-top box alarm data; the second category is user complaint data.

[0027] Such as figure 1 The sampling method flow shown, specifically including:

[0028](1) Initialization: Select 10 attributes of the data, and then clean the data. The main goal of cleaning is to remove irrelevant data and redundant information, namely noise samples and unusable data. Data cleaning includes the following two steps: 1. Clean up the wrong data, check the repeatability of the data and mark the samples. The processing of these data is conducive to improving the classification results and avoiding the over-generalization of the data set. 2. Traverse each sample in the entire complaint data set 1, and mark ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an improved SMOTE re-sampling method for unbalanced data classification. The method comprises clustering a minority class of samples in a sample set by using a K-Means method, deleting the noise sample class with a centroid closest to a majority of samples in each class cluster, classifying each class cluster into three classes by using a KNN method and removing the noise sample class, finally inputting a random number in each class cluster and selecting a certain sample set according to the proportion relation of the random number to the sample set type in the class cluster to carry out the SMOTE method oversampling. Compared with a traditional SMOTE method, the improved K-Means-SMOTE method is significantly improved in effect in a model of predicting the complaint of network television set-top box users.

Description

technical field [0001] The invention relates to an improved SMOTE re-sampling method for unbalanced data classification, belonging to the technical field of unbalanced data classification. Background technique [0002] In practical applications, the original data objects we get are often unbalanced, that is, the number of samples of a certain category is much larger than that of other categories, such as medical diagnosis, network intrusion, and IPTV prediction failure model. Among them, we call the class with a relatively large number as the majority class, and the class with a relatively small number is called the minority class. When traditional classifiers deal with unbalanced data, the classifiers usually trained are biased towards the majority class, that is, the prediction accuracy rate for the majority class is high, while the accuracy rate for the minority class is relatively low. At present, the processing methods for unbalanced datasets can generally be divided i...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62
CPCG06F18/23G06F18/23213
Inventor 周亮王堂辉魏昕刘榕华张胜男赵磊
Owner NANJING UNIV OF POSTS & TELECOMM
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap