Electrocardiograms

Download the dataset

The dataset is available here. Let's download it and store it in the data folder in the 02-Electrocardiograms directory with the following commands:

cd ~/code/<user.github_nickname>/{{local_path_to("05-ML/03-Performance-metrics/02-Electrocardiograms")}}
curl https://wagon-public-datasets.s3.amazonaws.com/Machine%20Learning%20Datasets/ML_Electrocardiograms_dataset.csv > data/electrocardiograms.csv

The dataset

• Each observation of the dataset is a numerically represented heartbeat, taken from a patient's electrocardiogram (ECG).
• The target is binary and defines whether the heartbeat is at risk of cardiovascular disease [1] or not [0].

Exercise

🎯 Your task is to flag heartbeats that are at risk of cardiovascular diseases. You will:

• Investigate the class balance of the dataset
• Evaluate and compare two models: KNN and LogisticRegression
• Use Confusion matrix and Classification report to get insight on the models' performances
• Choose the optimal model based on the appropriate metric

To start the exercise, open Electrocardiograms.ipynb in jupyter notebook and follow the instructions.

🚀 Your turn!