---

Heart Disease Prediction using Machine Learning

A machine learning project that predicts whether a patient has heart disease using five classification algorithms trained on the UCI Heart Disease Dataset (Cleveland Clinic Foundation).

• 0 = No heart disease present
• 1 = Heart disease present

---

Project Structure

├── create_csv.py                    # Script to generate the dataset file
├── heart_disease_prediction.py      # Main machine learning script
├── heart_disease_for_excel.csv      # Real-world dataset (1,025 records)
└── README.md                        # Project overview (this file)

---

Dataset

The project uses the real-world UCI Heart Disease Dataset. It contains clinical data from 1,025 patients with 13 clinical features and 1 target variable.

Feature	Description	Type
age	Age of the patient	Integer
sex	Sex (0 = female, 1 = male)	Binary
cp	Chest pain type (0–3)	Ordinal
trestbps	Resting blood pressure (mmHg)	Integer
chol	Serum cholesterol (mg/dl)	Integer
fbs	Fasting blood sugar > 120 mg/dl	Binary
restecg	Resting ECG results (0–2)	Ordinal
thalach	Maximum heart rate achieved	Integer
exang	Exercise-induced angina (0/1)	Binary
oldpeak	ST depression induced by exercise	Float
slope	Slope of peak exercise ST segment	Ordinal
ca	Number of major vessels (0–4)	Ordinal
thal	Thalassemia type (0–3)	Ordinal

Train/Test Split: 80% training (820 samples) / 20% testing (205 samples).

---

Workflow

1. Data Preprocessing & EDA

• Dataset Loading: Handled CSV metadata (sep=,) and binarized target values.
• Outlier Detection: Identified outliers in trestbps, chol, thalach, and oldpeak using the IQR method.
• Correlation Analysis: Identified cp (Chest Pain) and thalach (Max Heart Rate) as top positive predictors, while oldpeak and exang showed strong negative correlations.
• Feature Scaling: Applied StandardScaler to normalize the feature ranges.

2. Modeling & Evaluation

• Baseline Models: Logistic Regression and Naive Bayes.

• Advanced Models: - KNN: Tuned for the best K-value.

• SVM: Utilized a linear kernel.

• Random Forest: Ensemble method with depth constraints to prevent overfitting.

• Metrics: Evaluated using Accuracy, Precision, Recall, and F1-Score.

---

Model Performance Summary

Rank	Model	Accuracy	Precision	Recall	F1-Score
1	Random Forest	92.68%	0.8947	0.9714	0.9315
2	KNN	84.39%	0.8230	0.8857	0.8532
3	Naive Bayes	82.93%	0.8070	0.8762	0.8402
4	SVM	81.46%	0.7638	0.9238	0.8362
5	Logistic Regression	80.98%	0.7619	0.9143	0.8312

Best model: Random Forest — 92.68% accuracy

---

How to Run

1. Prepare the Data: Run the CSV creation script to generate the dataset:

python create_csv.py

2. Train and Evaluate: Run the main prediction script:

python heart_disease_prediction.py

Requirements:

pandas
numpy
scikit-learn

Install dependencies via pip:

pip install pandas numpy scikit-learn