Open AccessResearch on complaint prediction based on feature weighted Naive Bayes
Author(s) -
Zheng Jian,
Wei Yang,
Changchun Wang,
Dandan Jiang,
Dan Wang,
Qi Yang,
Yijiao Zhang
Publication year - 2022
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/983/1/012117
Subject(s) - naive bayes classifier , complaint , weighting , artificial intelligence , computer science , classifier (uml) , machine learning , bayes' theorem , bayes classifier , pattern recognition (psychology) , bayes error rate , data mining , support vector machine , bayesian probability , medicine , political science , law , radiology
Complaint prediction is a hot topic in current research. Due to complaint prediction is random, the accuracy of complaint prediction is not high. To solve this problem, this paper proposes a complaint prediction method based on feature weighted Naive Bayes. According to rough set theory, we first calculate the importance of each conditional attribute, combine the importance of attribute as the weight with the naive Bayes classifier to form a weighted naive Bayes classifier. We use multiple classifiers to make predictions, the latter classifier performs iterative learning on the basis of the previous classifier, and finally all classifiers are given different weights to make decisions. The experimental results show that the proposed method effectively combines the weighting method and the Naive Bayes method to achieve reliable prediction of complaints.