💊 Drug Prediction — Decision Tree

A clear and interpretable baseline for predicting drug categories using patient features and a Decision Tree classifier. Designed to be interview-friendly, with emphasis on clarity, step-by-step decisions, and interpretability.

---

📂 Project Structure

├── Drug_Prediction_DecisionTree_polished.ipynb
├── README.md # Project documentation

---

⚙️ Skills & Tech

• Python, Jupyter Notebook
• pandas, NumPy — Data handling
• matplotlib, seaborn — Visualization
• scikit-learn — DecisionTreeClassifier, model evaluation
• EDA, Preprocessing, Model interpretation

---

📝 Project Overview

This notebook demonstrates a complete Machine Learning workflow for predicting drug categories:

1. Exploratory Data Analysis (EDA) – Inspect dataset distribution and patterns
2. Preprocessing – Encoding categorical features, handling data types
3. Model Training – Decision Tree Classifier
4. Evaluation – Accuracy, interpretability, decision paths

---

📊 Dataset

• Features:
  • Age: Age of the patient
  • Sex: Male/Female
  • Blood Pressure: Low / Normal / High
  • Cholesterol: Normal / High
  • Na_to_K ratio: Sodium-to-Potassium ratio in the blood
• Target: Drug type (DrugA, DrugB, DrugC, DrugX, DrugY)
• Size: 200 samples
• Source: UCI / educational dataset

---

▶️ How to Run

1. Clone or download this repository:

   git clone https://github.com/Shamir-Havas/Drug-Prediction-Decision-Tree.git
   cd Drug-Prediction-Decision-Tree

Install dependencies:

bash Copy code pip install -r requirements.txt Open Jupyter Notebook and run the workflow:

bash Copy code jupyter notebook Drug_Prediction_DecisionTree_polished.ipynb Run all cells:

Kernel → Restart & Run All

📊 Results

🔹 Category Counts

🔹 Decision Tree Visualization

🔹 Model Accuracy

🔍 Model Explainability

Decision Tree Visualization: Interpretable decision paths using plot_tree

Classification Report: Precision, recall, F1-score

🚀 Future Improvements

Hyperparameter tuning with GridSearchCV / RandomizedSearchCV

Cross-validation (e.g., Stratified K-Fold) for robustness

Try ensemble methods (Random Forest, XGBoost)

Domain-specific validation & feature engineering

📦 Requirements

pandas==2.0.3
numpy==1.25.2
matplotlib==3.7.2
seaborn==0.12.2
scikit-learn==1.3.0