Closed lcarasik closed 6 years ago
I didn't transform the data at all. I just took the absolute value of the difference between the accepted value and the calculated value and divided by the accepted value. Then took the mean of this value across all the data points.
Need to address calculating mean percent error (or other metric) based on log transformed values as opposed to based on non-transformed. To help prevent higher percent errors (or other metric) from heavily biasing the mean value.