LandMapper new functionalities
The LandMapper
class provides a lot of functionalities to run the sckit-learn machine learning methods in raster data, however the follow improvements need to implemented:
-
Support for regression problems (regression branch) -
Correction factor for prediction uncertainty, according to mse derived by CV (R implementation and Medium article) -
Possibility to subset the training dataset, keeping the same distribution of classes of the full dataset, during the hyper-parameter optimization
Edited by Leandro Parente