RStoolbox 0.1.3

New:

• new logical argument predict for superClass. Disables prediction of full raster (proper validation is still conducted, but the prediction is done only for sample pixels).
• new generic predict() function for superClass objects. Useful to separate model training and prediction.
• new example data set (landcover training polygons) for lsat example data under /extdata/trainingPolygons.rds

Fixes:

• fix histMatch for single layers (affected also 'ihs' pan-sharpening)
• fix superClass validation sampling for factors (character based factor responses could lead to wrong factor conversions and wrong validation results)
• improved handling of of training polygons with overlaps and shared borders in superClass
• improved checks and error messages for insufficient training polygons