Using Data visualization/analysis, and machine learning models to build a model for detecting heart disease. Different machine learning models were trained and their performances were compared to find the best model. After finding out that Random Forest and XGBoost performs the best, now it is time to deploy the model.
For deploying I chose the XGBoost model. An xgboost deployment pipeline is better because retraining XGBoost is faster than random forest. First I created a S3 bucket to store the training data. Then I retrained the model using the same process and parameters that I used while building the model locally. Then I load and reformat the CSV data before deploying the trained model to an endpoint, finally running the model to make predictions.
- Dataset: https://www.kaggle.com/datasets/johnsmith88/heart-disease-dataset
- AWS documentations:
2.1) https://aws.amazon.com/getting-started/hands-on/build-train-deploy-machine-learning-model-sagemaker/
2.2) https://sagemaker.readthedocs.io/en/stable/frameworks/sklearn/using_sklearn.html#deploy-sklearn-models