I'm using the latest version of RStoolbox and R 3.6.3. When using superClass with model= "ranger", type="prob", and predict=TRUE, I sometimes get layers for each class that just contain 0 or 1 indicating whether or not a pixel was classified as that class rather than the probability. If I follow up and repredict using the fitted superclass object and the same data using predict.superClass, I get maps of the probabilities, so it seems to be an issue with the prediction within superClass rather than the model fitting. It doesn't happen every time, but it seems to be more common when I use beginCluster. I am setting the random seed so results should be consistent across runs.
I'm using the latest version of RStoolbox and R 3.6.3. When using superClass with model= "ranger", type="prob", and predict=TRUE, I sometimes get layers for each class that just contain 0 or 1 indicating whether or not a pixel was classified as that class rather than the probability. If I follow up and repredict using the fitted superclass object and the same data using predict.superClass, I get maps of the probabilities, so it seems to be an issue with the prediction within superClass rather than the model fitting. It doesn't happen every time, but it seems to be more common when I use beginCluster. I am setting the random seed so results should be consistent across runs.