To predict if a person has diabetes or not using decision tree with an accuracy of 77% By removing one feature I got an accuracy percentage of 76% I added both files differently
Describing the scope of this project diabetes is one of the major problem faced by many people in present world. But the main sorrow is that many might not even know if there are suffering from diabetes and by the time they realize it might be too late. So to prevent this I used a machine learning algorithm called Decision tree Classifier to predict if the person is suffering from diabetes or he might have a chance to get diabetes in future. The accuracy of my classifier is 78% but if we consider the case of might has the chance then the accuracy is around 91%.
Going into the technical decision the main reason for the selection of Decision tree is that its accuracy rate is pretty high compared to Random forest and KNN classifiers. I used the standard data set from kaggle and using that I learned what the effect of different factors like glucose percentage, insulin percentage etc., effects the or increase the chances of diabetes and with that knowledge and dataset present with me I have successfully solved the prediction.
"For further explanation see the following blog post written by me to understand the problem in much depth"
https://machirajublog.wordpress.com/2018/04/22/my-first-ml-project/