Skip to content

Compared the metrics and performance of different classification algorithms on Heart Failure dataset from UCI ML Repository

Notifications You must be signed in to change notification settings

grvnair/heart-failure-prediction

Repository files navigation

heart-failure-prediction

Dataset for this project is taken from UCI ML Repository: https://archive-beta.ics.uci.edu/dataset/519/heart+failure+clinical+records

This project is based on a binary classification task to identify if the patient is a high risk patient. I compared different classification algorithms on their Accuracy, Precision, Recall and F1 Score. I started by analysing the data, and then trained and tested the models on Logistic Regression, K-Nearest Neighbors, Support Vector Classifier, Decision Tree Classifier, and Random Forest Classifier. I used the Scikit-learn library to implement and evaluate each algorithm.

After testing all the algorithms, the best metrics achieved were Accuracy - 0.8, Precision - 0.88, Recall - 0.6 and F1 Score - 0.71.

After working on this project, I learned the importance of comparing and evaluating different algorithms across important metrics to get a robust and reliable model.