Diabetes Prediction Using Machine Learning

This project focuses on predicting diabetes using multiple machine learning classification algorithms. The dataset is analyzed, cleaned, visualized, and modeled using classical ML techniques.

---

📊 Dataset

• Source: Pima Indians Diabetes Dataset
• Features include:
  • Pregnancies
  • Glucose
  • BloodPressure
  • SkinThickness
  • Insulin
  • BMI
  • DiabetesPedigreeFunction
  • Age
• Target variable: Outcome (0: Non-diabetic, 1: Diabetic)

---

🔍 Exploratory Data Analysis (EDA)

• Dataset overview (info, describe)
• Pairwise feature relationships (Pairplot)
• Correlation heatmap
• Outlier detection using IQR method

---

🧹 Data Preprocessing

• Outlier removal with Interquartile Range (IQR)
• Train-test split (75% train / 25% test)
• Feature scaling using StandardScaler

---

🤖 Machine Learning Models Used

• Logistic Regression
• Decision Tree
• K-Nearest Neighbors (KNN)
• Naive Bayes
• Support Vector Machine (SVM)
• AdaBoost
• Gradient Boosting
• Random Forest

Model performance evaluated using 10-Fold Cross Validation.

---

⚙️ Hyperparameter Optimization

• GridSearchCV applied on Decision Tree
• Best parameters selected using 5-Fold CV

---

📈 Model Evaluation

• Confusion Matrix
• Classification Report (Precision, Recall, F1-score)
• Accuracy comparison using boxplots

---

🧪 Prediction on New Data

The trained model can predict diabetes for new patient data:

new_data = [[6,149,72,35,0,34.6,0.627,51]]