Dear All,can somebody kindly help me: i have to choose 2 independent sets in classifikation of land use as: Training sample set for classifikation procedure Testing sample set for accuracy assessment In "Generate Random Sample"-toogle i can define ROIs for training set (in my example should be equalised random sampling= 240 pixels for training set) and conduct any classifikation. But i do not understand, how should i make a testing set only for accuracy assessment (with other sample size=750) so that this set is undependent sample set from my training set. I do should use the testing set if i want to produce confusion matrix. For this purpose i take classification result file, but this file takes automatically ROIs from training set (=240 pixels), so that i really do not realize, how can i USE FOR CONFUSION MATRIX TESTING SET for postclassifikation, should i conduct new classification with new ROIs for testing set? Please help me...i need indeed as soon as possible any help, thanks a lot in advance!!!