Abstract:
Studies have shown that diabetes is becoming one of the major risk factors to people’s health.
Diabetes involve problems with the serum levels of glucose and insulin, but its inception could
be reflected by a series of other symptoms. This case aims at applying some Bayesian methods
to construct a Bayesian logistic regression model and a naive Bayes model. We want to classify
samples with and without diagnosed diabetes from predictive variables. Also, we compare their
performance of different models including traditional logistic regression model. Three models’ hit
rates are more than 0.7, and the hit rate of the naive Bayes model is less than the other models.
These classifiers may classify some healthy people as diabetes patients, but some patients might
be misdiagnosed as normal person.
Keywords: diabetes, logistic regression, Bayesian regression, naive Bayes
贝叶斯统计论文.rar
(204.49 KB, 需要: RMB 10 元)


雷达卡




京公网安备 11010802022788号







