🩺 Week 1 — Disease Prediction Using Patient Data

This project is part of the DevelopersHub Internship (AI/ML).
The task for Week 1 was to learn the basic ML workflow by predicting heart disease using patient data from the UCI Cleveland Heart Disease dataset.

---

📂 Project Structure

WEEK1-disease-prediction/ │── data/
│ └── cleveland.csv # Dataset (renamed from processed.cleveland.data)
│
│── notebooks/
│ ├── 01_load_and_explore.ipynb # Step 1: Load + Explore dataset
│ ├── 02_preprocessing.ipynb # Step 2: Preprocessing (imputation, scaling, binary target)
│ ├── 03_eda.ipynb # Step 3: Exploratory Data Analysis
│ ├── 04_model_training.ipynb # Step 4: Logistic Regression & Random Forest
│ └── 05_evaluation_report.ipynb # Step 5: One-page summary
│
│── week1_report.md # Short 1-page Markdown report
│── week1_report.pdf # Exported PDF report (submission)
│── README.md # Project documentation (this file)

---

📊 Dataset

• Source: UCI Machine Learning Repository (Cleveland subset, processed version)
• Size: 303 rows × 14 columns
• Target: target (0–4) → binarized to target_bin (0 = healthy, 1 = disease)

---

⚙️ Preprocessing

• Missing values:
  • ca, thal → filled with mode (most frequent value)
  • Other numeric columns → filled with median
• Feature scaling: All features scaled to [0, 1] using MinMaxScaler
• Final dataset: 13 features + 1 binary target (target_bin)

---

🔍 Exploratory Data Analysis (EDA)

• Class balance: ~54% healthy, ~46% disease
• Feature distributions: Plotted histograms for all features
• Correlation heatmap: Studied relationships between features and target

---

🤖 Models & Results

Two models were trained and evaluated:

Model	Accuracy
Logistic Regression	0.8525
Random Forest	0.9016 ✅

Selected Model: Random Forest (better accuracy)

• Why Random Forest? Random Forest outperformed Logistic Regression due to its ability to model complex interactions and non-linear relationships among features.

---

📄 Outcome

• Learned the complete ML workflow:
  • Data loading → preprocessing → EDA → model training → evaluation
• Produced a 1-page report (week1_report.pdf) for submission
• Random Forest performed better, and it was selected as the baseline model.

---