🩺 Diabetes Prediction App

A machine learning web application that predicts whether a patient is diabetic based on medical diagnostic measurements. Built with Python, Scikit-learn, and deployed using Streamlit.

---

🚀 Live Demo

👉 Click here to try the live app!

Deployed on Streamlit Cloud

---

📌 Problem Statement

Diabetes affects over 400 million people worldwide. Early detection is critical for treatment and prevention of complications. This app allows patients and doctors to quickly assess diabetes risk based on simple medical inputs.

---

🔧 Tech Stack

Tool	Purpose
Python	Core programming language
Pandas & NumPy	Data manipulation and analysis
Matplotlib & Seaborn	Data visualization and EDA
Scikit-learn	Machine learning model building
Streamlit	Web app deployment
Pickle	Model saving and loading

---

📊 Dataset

• Source: PIMA Indians Diabetes Dataset (UCI Machine Learning Repository)
• Records: 768 patients
• Features: 8 medical features (Glucose, BMI, Age, Insulin, Blood Pressure, etc.)
• Target: Diabetic (1) or Not Diabetic (0)

---

🧪 ML Pipeline

1. Exploratory Data Analysis (EDA) — Analyzed feature distributions and correlations
2. Data Preprocessing — Handled zero values (invalid medical readings), scaled features
3. Feature Selection — Identified most important predictors of diabetes
4. Model Building — Trained and evaluated:
   • Logistic Regression
5. Model Evaluation — Accuracy, Confusion Matrix, Classification Report
6. Deployment — Saved model as .pkl and built interactive Streamlit app

---

📈 Model Performance

Model	Accuracy
Logistic Regression	~78%

---

🖥️ How to Run Locally

# 1. Clone the repository
git clone https://github.com/nithinreddyp2004/diabetes-prediction.git

# 2. Navigate to the project folder
cd diabetes-prediction

# 3. Install dependencies
pip install -r requirements.txt

# 4. Run the Streamlit app
streamlit run app.py

---

📁 Project Structure

diabetes-prediction/
│
├── model_training.ipynb     # Jupyter notebook with EDA and model training
├── app.py                   # Streamlit web application
├── model_logistic.pkl       # Saved trained model
├── diabetes.csv             # Dataset
├── requirements.txt         # Python dependencies
└── README.md                # Project documentation

---

🔍 Key Insights from EDA

• Glucose level is the strongest predictor of diabetes
• Patients with higher BMI are at greater risk
• Age above 45 shows significantly higher diabetes rates
• Insulin levels show high variance — requires careful handling

---

🩺 Input Features Used

Feature	Description
Pregnancies	Number of times pregnant
Glucose	Plasma glucose concentration
Blood Pressure	Diastolic blood pressure (mm Hg)
Skin Thickness	Triceps skin fold thickness (mm)
Insulin	2-Hour serum insulin (mu U/ml)
BMI	Body mass index
Diabetes Pedigree Function	Diabetes likelihood based on family history
Age	Age in years

---

👨‍💻 Author

Nithin Reddy

• GitHub: @nithinreddyp2004

---

📄 License

This project is open source and available under the MIT License.