Predicting Diabetes Using Health & Socioeconomic Indicators

Tags: Machine Learning | Python | Logistic Regression | Scikit-learn

Link: GitHub

🔍 Overview

Used CDC’s BRFSS 2015 dataset to build a predictive model for early detection of diabetes based on lifestyle, demographic, and health metrics. The goal was to flag high-risk individuals before clinical onset.

❓ Key Questions

• Can diabetes risk be predicted using demographic and behavioral indicators?

• Which features most influence diabetes prediction?

🧪 Methodology

• Cleaned 250K records, removed duplicates, engineered key features

• Compared Logistic Regression, Decision Tree, and Quadratic classifiers

• Evaluated using F1-score, recall, and precision

📈 Results

• Best F1-score: 90.6% using Logistic Regression

• Key predictors: BMI, blood pressure, cholesterol levels