This repository contains a comprehensive analysis and prediction model focused on heart attack risks. The project utilizes Python libraries like NumPy, Pandas, Seaborn, and Matplotlib for data manipulation and visualization.
The heart.csv dataset includes various health metrics like age, sex, chest pain type, resting blood pressure, cholesterol levels, fasting blood sugar, resting electrocardiographic results, maximum heart rate, and others. It consists of 1025 entries with 14 columns, providing a rich dataset for analysis.
The EDA involves histogram plotting for age distribution, a heatmap for correlation analysis, and a detailed investigation of each feature's impact on heart attack risk.
We employ machine learning models like Linear Regression and XGBRegressor. The models are trained and evaluated using metrics like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R2 Score. A comparative analysis of these models is provided to determine the most effective approach.
The project explores Principal Component Analysis (PCA) for feature dimensionality reduction, assessing its impact on model performance.
We use K-Fold Cross-Validation for a more robust evaluation of model performance, providing insights into the generalizability of our models.
Run the "Heart Attack Predicton_EDA_Modeling" Notebook using the "heart.csv" data file.