The top 3 variables for each outcome in each model are the same, except for death where the logistic regression model indicated comorbidity count as a highly significant predictor, and where this predictor is ranked least important in the RF model, and sex was included in its place in the RF model, following a close 4 th in the logistic regression model.
← all excerpts
Using a cohort study of diabetes and peripheral artery disease to compare logistic regression and machine learning via random forest modeling.
1
—
—