🧠 Diabetes Prediction Using Machine Learning (CI512 Project)

A machine learning project using the Pima Indians Diabetes Dataset to predict whether a patient has diabetes. Built in 2021 for the CI512 – Intelligent Systems module, this project applies and compares multiple classification algorithms to identify the most accurate model.

---

📊 Dataset

• Source: National Institute of Diabetes and Digestive and Kidney Diseases
• Population: Female patients over 21 years old of Pima Indian heritage
• Features:
  • Pregnancies
  • Plasma glucose concentration
  • Diastolic blood pressure
  • Triceps skin fold thickness
  • Insulin
  • BMI
  • Diabetes Pedigree Function
  • Age
  • Outcome (0 or 1)

---

🤖 Algorithms Used

1. K-Nearest Neighbors (KNN)

   • Predicts based on the distance to nearest data points.
   • Chosen for its high accuracy with classification tasks.

2. Decision Tree Classifier

   • Easy-to-interpret model using rule-based branching.
   • Good for explaining predictions.

3. Random Forest Classifier

   • Ensemble method combining multiple decision trees.
   • Initially included, but later excluded due to inconsistent results.

4. Multilayer Perceptron (MLP)

   • Neural network model with hidden layers.
   • Provided the best prediction accuracy on this dataset.

5. Stacking Classifier

   • Combines multiple models to improve prediction using meta-learning.

---

🧪 Evaluation Method

• Data Cleaning: Replaced 0 values with feature-wise mean (to handle missing values).
• Validation: 10-fold cross-validation
• Split: 70% Training, 30% Testing
• Visualization: Line graphs for model accuracy comparison

---

📈 Results

• Random Forest was excluded due to poor alignment with other models.
• MLP (Multilayer Perceptron) outperformed all other classifiers.
• Stacking showed promise by combining model strengths.

---

🛠 Technologies

• Python
• Pandas
• Scikit-learn
• Matplotlib
• NumPy

---

👨‍💻 Developed By

Vickshan Vicknakumaran
University of Brighton
CI512 – Intelligent Systems (2021)

---