This project aims to facilitate early prediction of heart disease to aid in proactive healthcare management.
- Utilizes Heart Failure Prediction Dataset from Kaggle.
- Performs Exploratory Data Analysis (EDA) for initial insights.
- Builds a robust transformation pipeline for data preparation.
- Trains and evaluates various machine learning models using cross-validation.
- Deploys a user-friendly API using Flask for heart disease prediction.
Note: This project was made for educational purposes.
Python Version: 3.8
Packages: numpy, pandas, matplotlib, seaborn, scikit-learn, xgboost, flask, json, pickle
Setting Up Environment:
conda create -p venv python=3.8 -ypip install -r requirements.txt
Dataset: https://www.kaggle.com/datasets/fedesoriano/heart-failure-prediction
The project utilizes the Heart Failure Prediction Dataset, obtained from Kaggle. This dataset boasts a unique origin, combining five previously independent heart disease datasets that share 11 common features. Following the merging process, duplicate observations were removed, resulting in a final dataset of 918 observations.
The EDA examined data distributions and value counts for categorical variables. Key insights from pivot tables are visualized in the following figures:
- Split the data into train and test sets with a test size of 20%
- Constructed a data transformation pipeline that encodes categorical features and standardizes numerical features.
- Leveraging cross-validation, we evaluated multiple models to predict heart disease, prioritizing both accuracy and training efficiency. Ultimately, the Random Forest model emerged as the optimal choice due to its superior performance.
- Fine-tune the Random Forest model to achieve optimal performance.
The results of cross-validation for the models are as follows:
Deployed a user-friendly API using Flask. The API endpoint accepts user requests and returns the predicted heart disease classification.
