Skip to main content
. 2017 Jun 15;7:3582. doi: 10.1038/s41598-017-02303-0

Figure 4.

Figure 4

Classifier performance for Random Forests built on reaction fingerprints. Errors – both for (a) yields and (b) times – do not significantly decrease for large datasets. Blue, shaded regions demarcate lower and upper estimates for the Bayes classifier error.