Logistic Regression and Decision Tree ML Algorithms to Predict Type-2 Diabetes

Overview

Objectives/Goals Compare two Statistical Models to predict Type-2 Diabetes - Logistic Regression and Decision Trees. Determine which patient attributes - Age, Body Mass Index, Glucose Concentration, Genetics, % of time pregnant are most significant for Diabetes Determine the following for each model to aid comparison: Accuracy, Sensitivity, Specivity, ROC Area under Curve. Build a simple web application to use the model in mobile phones. Application should accept key patient data and return probability of diabetes Application should run on phone and browser. Methods/Materials UC Irvine Department of Machine Learning Pima Indians Diabetes DataSet. This dataset provides details on 782 Pima Indians for Age, BMI, Pregnancy etc. Scikit-learn: Machine learning in Python Logistic Regression and Decision Tree algorithm packages in Python. Pythonanywhere for Hosting and running Python Applications. Jupyter notebooks running on Azure Cloud. Methods Scikit-learn Machine Learning toolkit in Python was used for running Classification Models DataSet has 768 patient records which were divided into 75% (576 records) for Training data and remaining 25% (192 records) for Test data. Both models Logistic Regression and Decision Trees, are Trained and Scored with training data and test data respectively Prediction Accuracy is measured as (TP+TN) / (TP+TN+FP+FN) Sensitivity is measured as TP / (TP+TN) Specificity is measured as TN / (TN+FP) HTML5 was used to build a simple webapp that accepts Patient Data in a Form and calls backend Python App. Results Logistic Regression Model has

Competition history

• CSEF 2018 Computational Systems & Analysis · Entry S0827

Resources

• CSEF project summary ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Systems Software and Computer Science projects
• CSEF Systems Software and Computer Science projects
• Every CSEF 2018 project