🫀 Heart Disease Prediction: A Decision Tree Approach

---

📌 Project Overview

Cardiovascular diseases (CVDs) are the leading cause of death globally. Early detection is critical for effective treatment.

This project utilizes Machine Learning to build a diagnostic tool capable of predicting the presence of heart disease based on patient medical attributes. By employing a Decision Tree Classifier and optimizing it via Hyperparameter Tuning, this model provides interpretable and accurate predictions.

Objectives:

• Analyze clinical data to identify key risk factors.
• Build a predictive model to classify patients as "Healthy" or "Disease Detected".
• Visualize the decision-making process of the algorithm.

---

📂 The Dataset

The model is trained on the Cleveland Heart Disease Dataset (UCI Machine Learning Repository). It consists of 303 patient records with 14 distinct features.

Attribute	Description	Type
age	Age in years	Numerical
sex	1 = Male; 0 = Female	Categorical
cp	Chest Pain Type (0: Typical Angina, 1: Atypical, 2: Non-anginal, 3: Asymptomatic)	Categorical
trestbps	Resting Blood Pressure (mm Hg)	Numerical
chol	Serum Cholestoral in mg/dl	Numerical
fbs	Fasting Blood Sugar > 120 mg/dl (1 = True)	Categorical
restecg	Resting ECG results (0, 1, 2)	Categorical
thalach	Maximum Heart Rate Achieved	Numerical
exang	Exercise Induced Angina (1 = Yes)	Categorical
oldpeak	ST depression induced by exercise relative to rest	Numerical
slope	Slope of the peak exercise ST segment	Categorical
ca	Number of major vessels (0-3) colored by fluoroscopy	Numerical
thal	Thalassemia (1: Fixed defect, 2: Normal, 3: Reversible defect)	Categorical
target	Diagnosis (1 = Disease, 0 = No Disease)	Target

---

🛠️ Technologies Used

• Pandas: Data manipulation and cleaning.

• NumPy: Numerical operations.

• Matplotlib / Seaborn: Heatmaps, Bar charts, and Tree visualization.

• Scikit-Learn: Machine Learning implementation (DecisionTree, GridSearchCV).

  ---

⚙️ Methodology

1. Data Preprocessing

• Cleaning: Checked for null values and removed duplicate entries to prevent model bias.
• Encoding: Applied One-Hot Encoding to nominal categorical variables (cp, restecg, thal) to ensure the model treats them mathematically correctly without assuming a hierarchy.

2. Model Selection

• Algorithm: Decision Tree Classifier.
• Why? Decision Trees are highly interpretable and mimic human decision-making logic.

3. Hyperparameter Tuning

• To prevent Overfitting (where the model memorizes the data instead of learning patterns), I used GridSearchCV.
• Optimized Parameters:
  • max_depth: Controlled the vertical growth of the tree.
  • min_samples_split: Ensured nodes have enough data before splitting.
  • criterion: Tested both 'Gini Impurity' and 'Entropy'.

---

📊 Model Performance

The model was evaluated on a held-out Test Set (30% of data).

• Accuracy: ~81.32%

• Precision: High precision indicates a low False Positive rate.

• Recall: High recall indicates a low False Negative rate (crucial in healthcare).

---

🔍 Key Findings

Using Feature Importance analysis, the model identified the following as the strongest predictors of heart disease:

• Chest Pain Type (cp): Specifically asymptomatic pain.

• Thalassemia (thal): Reversible defects were highly correlated with disease.

• Number of Major Vessels (ca): Patients with fewer visible vessels often had higher risk.

• Max Heart Rate (thalach): Lower max rates during stress tests indicated issues.