CM-02Product sheet
Compare
One model, many places. Compare sets two runs side by side and shows how the models' performance holds, or does not, from one area to another.
One model, many places
Compare sets two runs side by side: two areas in the same region, or in different states. For each, it shows the confusion against the reference. Between them, it shows the accuracy delta overall and per class, the change in class shares, and agreement block by block.
Agreement, cell by cell
Agreement with MapBiomas is computed cell by cell, with producer's and user's accuracy for each class and Wilson intervals around both.
Quantity is not allocation
Quantity error and allocation error are reported apart, because getting how much soybean there is wrong is a different failure from getting where it is wrong.
An average hides a corner
Agreement is also broken into blocks across space. One number over the whole area hides whether the disagreement sits in one corner or throughout it, and disagreement throughout usually means the model is being asked about ground it did not learn. That question belongs to Diagnose.
!Current constraints
- MapBiomas is not field truth. Agreement is concordance with an annual map, often offset in time from the Sentinel-2 series.
- The accuracy assessment is not included in the research pack.