Probability Map changes for better usability and correct user input calculation
Due to #20 (closed) I suggest changes in the way how the user gets presented the outcome of how many models predict a pixel to a certain class and how the threshold is entered and processed further to make sure the outcome is what the user intended.
- The user gets presented the absolute number of models, not the relative
- The user can type in the absolute number of models as threshold
- The user's input is processed correct and leads to the intended assignment
- Maybe a new name for the "probability map" would be nice, because now there aren't shown probabilities anymore
The following tasks need to be done for HaSa:
-
in
plot_interactive.r -
change the calculation of the raster image presented to the user -
change the legend of the map for threshold selection -
in
outer_procedure.r -
change the calculation of the threshold
After the changes were applied the assignment of pixels related to the threshold input is correct now every time.
The adjustment of the legend of the (probability) map was kind of tricky, because the values vary now. That's why there is a case distinction. The reason for that is, if there is a small number of models the legend contained decimal steps (which is wrong, because there are no "half models"), so whole number steps were implemented. That in turn led to very large legends if there is a large number of models.
Now the legend has a maximum of 10 entries.
The only small problem remaining is, that the maximum value (max. number of models) does not always appear in the legend due to automated creation. The max. number of models is printed out above the map, but it still could confuse unexperienced users.