Summary of results from ensemble models
Summary of results from ensemble models#
Data used |
Logistic regression |
Random forests |
Neural network (fully connected) |
---|---|---|---|
Use original patient data only |
83.2% |
84.2% |
83.4% |
Model probability output only |
83.6% |
83.8% |
83.5% |
Original patient data + model probability output |
83.0% |
84.2% |
83.0% |
Use of ensemble techniques did not improve model accuracy.