search for: traist

...n was correct. Now i am interested in predicting the threat level of a species not purely as threatened or not but to specific IUCN levels. I have used the predict.lrm function (predict.lrm(model1, type="fitted.ind")) to generate probabilities for each level, and also (predict(model1, traist, type="fitted")) see below. When i call the model the Brier score is 0.201 and C value 0.677. However when i inspect the output and relate it to the corresponding species ( for which i know the true IUCN rating) the model performs very badly, only getting 43% correct. Interestingly i...