Heart Disease Analysis and Prediction

This workflow demonstrates data exploration, visualization, and classification model training for heart disease prediction. Below is an outline of the steps taken:

---

1. Data Loading and Inspection

1. Load the dataset: A CSV file named heart.csv is read into a pandas DataFrame.
2. Initial exploration: Display the first few rows (head()), view descriptive statistics (describe()), and confirm there are no missing values (isnull().values.any()).
3. Basic checks: Investigate the distribution of specific features (e.g., sex).

---

2. Exploratory Data Analysis (EDA)

1. Subset the data: Separate records for patients with heart disease (output == 1) and those without (output == 0).
2. Histograms: Plot the distribution of features like age and thalachh (max heart rate) for each group.
3. Boxplots: Compare the distributions of numeric features (e.g., thalachh) between the two groups via boxplots.
4. Correlation: Generate a correlation matrix to identify relationships between variables. Visualize it using heatmaps (e.g., Seaborn, Plotly).

---

3. Visualizations

1. Seaborn plots:
   • Violin plots, pairplots, strip plots to understand data patterns and potential class separability.
2. Plotly:
   • Interactive histograms and boxplots (e.g., for output, colored by sex) provide insights into how different demographics relate to heart disease.
   • Boxplots for features like age, thalachh, and trtbps grouped by output.

---

4. Data Preprocessing

1. Information check: Review data types and non-null counts (df.info()).
2. Splitting:
   • Assign the target variable (output) to y and the remaining columns to X.
   • Perform a train–test split (e.g., 80/20) to separate the dataset for training and validation.
3. Scaling:
   • Use a standard scaler (StandardScaler) to normalize numeric features in both training and test sets.

---

5. Model Training

1. Define classifiers:
   • Logistic Regression
   • Decision Tree
   • Random Forest
   • K-Nearest Neighbors
   • Support Vector Classifier
   • XGBoost
2. Fitting: Train each model on the scaled training data.
3. Predictions: Generate predictions on the test set.
4. Evaluation: Calculate accuracy scores (accuracy_score) to compare performance across models.

---

6. Hyperparameter Tuning

1. Parameter grid: Define candidate hyperparameters (e.g., n_neighbors, weights, metric for KNN).
2. GridSearchCV: Perform a grid or randomized search across these parameters to find the best combination.
3. Refit model: Using the best parameters, retrain and evaluate on the test set to check for improvements.

---

7. Results and Observations

1. Accuracy comparison: Identify the top-performing classifiers for heart disease prediction.
2. Visual insights: From EDA plots, note key differentiating features (e.g., thalachh and age across different output groups).
3. Refinement: Use hyperparameter tuning results to optimize models like KNN for better performance.

---

License

This project is open-source. You are free to use, modify, and distribute the code. If you use it in your own work, a reference or link back to this repository is appreciated.

---

Feel free to adapt this single-notebook guide to your preferences or add deeper domain-specific analysis.