Earlier Detection of Diabetes using Machine Learning

Abstract

In worldwide group of people experience the illness and suffer due to the excess blood sugar in diabetes cause complications. It can severely damage to vital parts of the body like kidney, eyes and cardio vascular and it may have the risk of heart attack and stroke. Early detection of Diabetes plays a very important key role in healthcare. Our system is to efficiently identify the presence of blood sugar level with checking the number of parameters and by applying the machine learning techniques. Diabetes identification is mainly based on applying the classification algorithm may includes Logistic Regression, Naive Bayes, SVM, Decision Tree, Random Forest with by use feature selection algorithm which is used to solve the feature selection problem. The feature selection algorithm which is accomplished of removing the unwanted variable which is having low variance and also it is used to reduce the irrelevant and redundant features [1]. The tentative outcome show that the comparison of classifier algorithm with Random Forest, Logistic regression, SVM, Naive Bayes, , Decision tree and prove that the Logistic regression gives the good accuracy level and further the algorithm applied with the feature selection dataset which refers removal of low variance data in the dataset by finding the correlation between the variable in the dataset which gives the good accuracy level compared with the dataset which is not the process of selecting a subset of relevant features for use in model construction it means we are considering the all field in the dataset