An end-to-end learning project: explore clinical features, compare classifiers, and serve the selected pipeline through Streamlit.

The notebook visualizes relationships between heart-disease labels and age, sex, chest-pain type, cholesterol, and other measurements. It encodes categorical values, standardizes numerical features, and compares Logistic Regression, Gaussian Naive Bayes, Decision Tree, SVM, and K-Nearest Neighbors. The Streamlit app loads the committed KNN model, scaler, and feature schema for interactive predictions.

Flow

flowchart LR
  CSV["heart.csv"] --> EDA["EDA + preprocessing"]
  EDA --> SPLIT["train/test split"]
  SPLIT --> MODELS["5 classifier comparison"]
  MODELS --> ART["model + scaler + columns"]
  ART --> APP["Streamlit app"]

Loading

Run the app

git clone https://github.com/ReaperXD67/heart-eda.git
cd heart-eda
python -m venv .venv
pip install -r requirements.txt
streamlit run app.py

Open Untitled.ipynb to reproduce the EDA and model comparison.

Repository map

Artifact	Role
heart.csv	Input dataset
Untitled.ipynb	EDA, preprocessing, and model comparison
knn_model.pkl	Serialized classifier used by the app
scaler.pkl	Serialized numerical transformer
model_columns.pkl	Expected inference feature order
app.py	Streamlit inference UI

Medical safety

Caution

This project is educational and is not a medical device. A prediction is not a diagnosis and must not be used to delay or replace professional care. The repository does not provide external clinical validation, calibration analysis, subgroup fairness evaluation, or deployment monitoring.