Cardiovascular Health Predictor

Project Overview

Dataset Information

The dataset originates from a reputable medical source and comprises multiple features significant to cardiovascular health, such as age, gender, blood pressure, cholesterol levels, and lifestyle choices. The data is preprocessed and transformed to ensure optimal model performance.

Analytical Approach

The project follows a structured approach:

1. Data Preprocessing: Cleaning, normalizing, and transforming data to prepare for analysis.
2. Exploratory Analysis: Uncovering trends and patterns through statistical methods and visualization.
3. Model Building: Implementing machine learning algorithms including Random Forest, AdaBoost, and XGBoost.
4. Evaluation: Assessing models using precision, recall, F1-score, and ROC-AUC metrics.

Configuration and Setup

The repository includes a Jupyter notebook with detailed code annotations. Key configurations:

• Python Version: 3.9
• Key Libraries: scikit-learn, pandas, numpy, matplotlib, seaborn, xgboost

Data Loading and Initial Exploration

The notebook begins with data loading, followed by an initial examination of the dataset's characteristics.

Data Cleaning and Preprocessing

Data quality assessment and transformations are meticulously performed to enhance model reliability.

Feature Engineering

Strategic feature engineering is applied, including BMI calculation and interaction terms to improve predictive power.

Exploratory Data Analysis (EDA)

An in-depth EDA is conducted, offering visual and statistical insights into the dataset, enhancing understanding of the underlying health factors.

Model Building and Evaluation

The project demonstrates the application of advanced machine learning models:

• Random Forest
• AdaBoost
• XGBoost

Each model undergoes thorough tuning and evaluation, ensuring robustness and accuracy.

Advanced Statistical Analyses

SHAP (SHapley Additive exPlanations) values are calculated for an in-depth understanding of feature importance.