Breast Cancer Predictor — Deep Neural Network (NumPy)

This project implements a Deep Neural Network from scratch using NumPy to classify Breast Cancer tumors as Benign or Malignant.

Unlike common ML workflows, this project does not use TensorFlow or PyTorch — every step including forward pass, backpropagation, and weight updates is implemented manually.

🚀 Features

• 5-Layer Neural Network (Hidden Layer Size = 8)
• Sigmoid Activation + Binary Cross Entropy Loss
• Manual Backpropagation & Gradient Descent
• Train-Test Split with Stratification
• Feature Scaling using StandardScaler
• Model Evaluation:
  • Accuracy
  • Precision
  • Recall
  • F1-Score
  • Confusion Matrix
• User-Input Prediction Mode

🧠 Model Architecture

Input: 30 features
Hidden Layers: 5 × 8 neurons
Output: 1 neuron (Sigmoid)

📊 Evaluation Results

Achieved 100% accuracy on test data (Confusion Matrix shows zero misclassification)

🛠 Tech Stack

Python, NumPy, Pandas, Scikit-Learn

🎯 Learning Outcomes

• Understood neural networks at mathematical level
• Implemented gradient flow manually
• Learned training vs inference behavior