#Diabetes Prediction using Binary Classification (Machine Learning)

Problem Statement

This project focuses on building a machine learning classification model to predict diabetes based on patient’s medical attributes. The objective is to analyze key factors involved to diagnose diabetes in an individual.

Problem Context

Diabetes is a major healthcare concern worldwide. Early prediction enables preventive care, timely diagnosis, and better treatment outcomes. This project demonstrates how data-driven models can support clinical decision-making.

Dataset

The dataset contains diagnostic health measurements from a publicly available dataset.

Source: Kaggle – Pima Indians Diabetes Dataset

Input Features:

• Pregnancies
• Glucose
• Blood Pressure
• Skin Thickness
• Insulin
• Body Mass Index (BMI)
• Diabetes Pedigree Function
• Age

Target Variable:

• Outcome (0 = Non - Diabetic, 1 = Diabetic)

[Dataset Features] (images./Dataset_features_diabetes.png)

Notebook (Google Colab)

This project notebook can be executed directly in Google Colab without any local setup. [Open in Google Colab] https://colab.research.google.com/drive/1r_wWsz3Q8Y12vikfxS2FUR4PPOtSfRTj?usp=sharing

Technical Approach

• Data cleaning and visualization
• Feature engineering
• Model training and comparison
• Model evaluation using classification metrics
• Usage of Binary Classification as supervised learning model where the target variable has exactly two possible classes.

Tech Stack

Python, Pandas, Scikit-learn, Seaborn, Matplotlib

Machine Learning Models

• Logistic Regression
• Decision Tree
• Random Forest Classifier
• K-Nearest Neighbors (KNN)
• Naive Bayes Model
• Ensemble Methods (Gradient Boosting and Ada Boost Model)

Model Evaluation

Evaluation was performed using the following metrics:

• Precision
• Accuracy
• Recall
• F1-Score
• ROC-AUC Score

Results

• Achieved accuracy of 76.62% on test data.

• Logistic Regression achieved a ROC AUC of 0.73.

  [ROC Curve] (images./ROC_curve_diabetes_dataset.png)

• Random Forest Classifier demonstrated superior predictive performance compared to other models.

  [Performance Scores]:

  {Before Preprocessing} (images./Performance_scores_before_preprocessing.png)

  {After Preprocessing} (images./Performance_scores_after_preprocessing.png)

Project Structure

diabetes-prediction/ ├── data/ │ └── diabetes.csv # Dataset ├── notebooks/ │ └── diabetes_prediction.ipynb # Exploratory analysis & modeling ├── src/ ├── preprocessing.py # Data cleaning, scaling, encoding ├── train.py # Model training logic └── evaluate.py # Metrics, ROC AUC, plots ├── requirements.txt # Dependencies └── README.md # Project documentation

Setup & Run

1. Clone the repository:

   git clone https://github.com/your-username/diabetes-prediction.git
   cd diabetes-prediction

2. Create and activate a virtual environment to isolate project dependencies. python -m venv venv source venv/bin/activate #macOS/Linux venv\Scripts\activate # Windows
3. Install dependencies: pip install -r requirements.txt
4. Run the notebook: jupyter notebook
5. (Recommended) Run on Google Colab by uploading the notebook/clicking on ‘Open on Colab’ button.

Key Learnings

• Gained insights into the importance of data scaling, as model performance varied significantly before and after preprocessing.

• Utilized correlation matrices to examine relationships among features.

  [Correlation Matrix] (images./Heatmap_diabetes_dataset.png)

  [Correlation Matrix] (images./Pairplot_diabetes_dataset.png)

• Acquired practical experience processing and analyzing real-world healthcare data.

Disclaimer

This project is developed for educational purposes using a publicly available dataset and is not intended for clinical or medical diagnosis.