Cancer Mortality Prediction

This project predicts cancer mortality rates using both Linear Regression and Deep Neural Networks (DNNs). It includes data preprocessing, model training, evaluation, and saving the best performing models for later testing.

---

📂 Project Structure

project/
│── preprocessor.py          # Data loading and preprocessing
│── linearRegression.py      # Linear regression model wrapper (SGDRegressor)
│── DNN.py                   # DNN model wrapper (Keras Sequential API)
│── utility.py               # Evaluation metrics and plotting functions
│── train_models.py          # Main script to train and save models
│── test_models.py           # Script to load and test saved models
│── cancer_reg-1.csv         # Dataset
│── saved_models/            # Folder where best models are saved
│   ├── best_linear_model.pkl
│   └── best_dnn_model.h5
│── plots/                   # Performance plots (generated automatically)
│── model_performance_summary.txt # Training summary results
│── README.md                # Project documentation

---

⚙️ Requirements

We are using an anaconda environment

Create the environment:

conda env create -f environment.yaml

Activate it:

conda activate dl-hw1

---

🚀 How to Run

1. Train Models

Run the training script to train Linear Regression and multiple DNNs with different learning rates. The script will:

• Train models.
• Evaluate performance (MSE, R²).
• Save plots in plots/.
• Save the best Linear Regression and best DNN models in saved_models/.

python train_models.py

After training, you’ll see:

• saved_models/best_linear_model.pkl
• saved_models/best_dnn_model.h5
• model_performance_summary.txt (training summary).

---

2. Test Saved Models

Run the testing script to load the saved models and evaluate them on the test set.

python test_models.py

This will print metrics and (optionally) generate performance plots for predictions.

Example output:

--- Test Results (LINEAR Model) ---
Model Path: saved_models/best_linear_model.pkl
Mean Squared Error: 240.1287
R^2 Score: 0.6543

--- Test Results (DNN Model) ---
Model Path: saved_models/best_dnn_model.h5
Mean Squared Error: 198.5632
R^2 Score: 0.7012

---

📊 Results

• Both Linear Regression and DNNs are trained.
• Best performing models are automatically saved.
• Performance plots are stored in the plots/ folder.

---