RCCS output
RCCS values are returned in the confidenceStats
array within "bins" of the following types:
default
- This type of bin has 18 bins that are evenly-spaced, except between bins 1→2 and 18→19 which are smaller. These bins represent the distribution of roof pixels across the dimension of 'not damaged' [top] to 'damaged' [bottom]. Those pixels scoring at the top of the bin column had decreasing evidence of damage, and those pixels scoring at the bottom of the bin column had increasing evidence of damage.extreme
- This type of bin contains three bins with the middle bin being by far the largest: 0.9921875. Hence this middle bin represents the percentage of roof pixels where confidence is not at either extreme; extremes represented by bin 1→2 and 3→4.
Below these, the summaries
array is displayed with the following details:
ratio
- This is the percentage of roof pixels with greater than near zero confidence that damage is present. median
- This is the confidence score where 50% of pixels have a lower, and 50% of pixels have a higher, confidence score.
IMPORTANT: Even though a higher confidence of damage indicates a higher level of damage, the relationship is not measurable.
Interpreting the output
The following illustration provides a guide to the RCCS and information on how to interpret different scores in the response.