search

Envirometrix / landmap

Landmap package for R
46 stars 13 forks source link

landmap do not calculate the error (logloss) for a binary response variable model #6

Open ManuelSpinola opened 3 years ago

ManuelSpinola commented 3 years ago

I am using landmap for a species distribution model with a binary variable 0 and 1 (as a factor).

There is not values for the error (logloss) .

This is how I fit the model:

tolomuco_landmap_binario <- train.spLearner(tolomuco_sp_binario["pa"], covariates = bio_cr_5367_grid[,c("bio1_23", "bio2_23", "bio3_23", "bio4_23", "bio5_23", "bio6_23", "bio7_23" , "bio8_23", "bio9_23", "bio10_23", "bio11_23", "bio12_23", "bio13_23", "bio14_23", "bio15_23", "bio16_23", "bio17_23", "bio18_23", "bio19_23", "alt_23")])

The results of the prediction:

head(pred_tolomuco_landmap_binario$pred) prob.0 prob.1 response error.0 error.1 1 0.6728527 0.3271473 0 NA NA 2 0.7445292 0.2554708 0 NA NA 3 0.7923370 0.2076630 0 NA NA 4 0.7866178 0.2133822 0 NA NA 5 0.8147125 0.1852875 0 NA NA 6 0.8105866 0.1894134 0 NA NA