Diabetes Classification

Problem

Predicting whether a patient has diabetes based on medical measurements.

Dataset

• Source: Pima Indians Diabetes Dataset
• Features: Glucose, BMI, Age, Blood Pressure, Insulin, etc.
• Target: Outcome (0 = No Diabetes, 1 = Diabetes)

Models Used

• Random Forest
• K-Nearest Neighbors
• Logistic Regression
• XGBoost
• SVC
• LinearSVC

Results

Model	Accuracy
Random Forest	0.753
KNN	0.701
Logistic Regression	0.759
XGBoost	0.740
SVC	0.772
LinearSVC	0.740

Best Model: SVC (Accuracy = 0.772)

What I Learned

• Comparing 6 classification models in one pipeline
• Tracking best model automatically
• Hyperparameter tuning with RandomizedSearchCV then GridSearchCV
• Visualizing confusion matrix for best model
• Using PolynomialFeatures with classification problems

Libraries

pandas, numpy, matplotlib, seaborn, scikit-learn, xgboost