Clara Wajdenbaum
The goal is to make predictions about people who have strokes based on the data provided.
healthcare-dataset-stroke-data.csv | https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset
I removed the rows with missing values and I removed the 'id' column. I scaled the integer and float columns and I have OneHotEncode the object columns.
No Stroke or Stroke by Age
Histogram
The Classification Tree model with 'max_depth' of 5, 'min_samples_leaf' of 20 and 'min_samples_split' of 2 has an accuracy score of 94.87%. The Logistic Regression Tree model has an accuracy score of 94.87%. The KNN model with kneighborsclassifier__n_neighbors'of 6, 'kneighborsclassifier__p' of 1 and 'kneighborsclassifier__weights' of 'uniform' has a score of 94.71%.
All three of our models work very well on our test set but I will choose the Classification Tree as I think it is the easiest one to use and also has the best score.
For any additional questions, please contact clara.wajdenbaum@icloud.com