Diabetes Classification Using Decision Tree

Project Overview

This project focuses on predicting diabetes in patients using a Decision Tree Classifier. The model is trained on medical diagnostic data to classify whether a patient is diabetic or non-diabetic. The project demonstrates a complete machine learning pipeline from data loading to model evaluation and visualization.

---

Dataset

The dataset consists of medical attributes commonly used for diabetes diagnosis.

Features

• Pregnancies
• Glucose
• BloodPressure
• SkinThickness
• Insulin
• BMI
• DiabetesPedigreeFunction
• Age

Target

• Outcome
  • 0 → Non-Diabetic
  • 1 → Diabetic

---

Technologies Used

• Python
• Pandas
• NumPy
• Matplotlib
• Seaborn
• Scikit-learn
• Graphviz
• Google Colab

---

Workflow

1. Load dataset using Pandas
2. Perform basic data inspection
3. Split dataset into training and testing sets (80:20)
4. Train Decision Tree Classifier
5. Evaluate model accuracy
6. Visualize decision tree

---

Model Training

from sklearn.tree import DecisionTreeClassifier

Model Evaluation

y_pred = model.predict(x_test)
accuracy_score(y_test, y_pred)

Accuracy

74.67%

Decision Tree Visualization

import graphviz

graphviz.Source(export_graphviz(
    model,
    feature_names=x.columns,
    filled=True
))

Project Structure

├── diabetes.csv ├── diabetes_decision_tree.ipynb ├── README.md

Future Improvements

Handle zero values using imputation Hyperparameter tuning Feature selection Compare with other models (SVM, Random Forest) Model deployment using Flask or FastAPI

Author

Saravanavel E AI & Data Science Student GitHub: https://github.com/SaravanavelE

License

This project is intended for educational and academic use.

model = DecisionTreeClassifier() model.fit(x_train, y_train)