Table 2 Machine Learning parameter values (p= the number of input variables)
From: A comparison of machine learning and Bayesian modelling for molecular serotyping
Parameter name | Step 1 training | Step 2 training | Package defaults |
|---|---|---|---|
GBM | Â | Â | Â |
Ntrees | 50 | 50 | 50 |
Learn_rate | 0.1 | 0.1 | 0.1 |
Max_depth | 25 | 5 | 5 |
Min_rows | 100 | 10 | 10 |
Balance_classes | FALSE | FALSE | FALSE |
Random Forest | Â | Â | Â |
Ntrees | 5000 | 100 | 50 |
Stopping_rounds | 2000 | Â | Â |
Max_depth | 100 | 20 | 20 |
Stopping_tolerance | 1e-20 | Â | Â |
Mtries | 100 | 200 | \(\sqrt {p}=21\) |
Balance_classes | TRUE | FALSE | FALSE |