Physiological Factors Affecting Diabetes in Women using Binary Logit Regression Model

Abstract

The purpose is to use statistical methods to analysis some female data, trying to figure out what physiological factors will affect the occurrence of diabetes on female and how. Method: using dataset from “AKSHAY DATTATRAY KHARE” in Kaggle, which comes from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) in original. In particular, all patients here are females of Pima Indian descent who are at least 21 years old. Use binary logit regression model to analysis. Result: Pregnancies, skin thickness, insulin and blood pressure, as long as age can’t be taken into account (p>0.05). The remaining three elements, which is glucose (OR=1.039), BMI(OR=1.073), diabetes pedigree function (OR=3.130), are all risk factors. The overall correct rate is 78.3%. The conclusion is using binary logit regression model can analysis factors related diabetes well. The logistic regression model among the influencing factors of diabetes shows that glucose, BMI, and diabetes pedigree function are likely to cause diabetes, and diabetes pedigree function has a greater impact on the psychological function of patients.