Options
Nash–Sutcliffe model efficiency coefficient
Contributor II
Hello.
I would like to know, if I can use in RapidMiner my own formula for evaluating results. In hydrology exist so called Nash–Sutcliffe model efficiency coefficient (http://en.wikipedia.org/wiki/Nash–Sutcliffe_model_efficiency_coefficient) . I do not fully understand user-based Performance - is it meant for this? Or maybe some combination of predicted and real labels can be evaluated? (by Generating attributes?)
Thanks,
Milan
I would like to know, if I can use in RapidMiner my own formula for evaluating results. In hydrology exist so called Nash–Sutcliffe model efficiency coefficient (http://en.wikipedia.org/wiki/Nash–Sutcliffe_model_efficiency_coefficient) . I do not fully understand user-based Performance - is it meant for this? Or maybe some combination of predicted and real labels can be evaluated? (by Generating attributes?)
Thanks,
Milan
0
Answers
I would use "generate attributes" to do the calculation. There would be a need to use macros. To get the result into a performance vector you could log the calculated result, turn this into another single row example set using "log to data" and then use "extract performance" to read this value.
regards
Andrew