Forums » Open discussion »
Feature Request: Gamma fit for percentile calculation
Added by Adrian Tompkins about 1 year ago
Hello, it's me again with another feature request regarding Percentile calculations.
Despite the huge choice of methods for percentile calculations, I was wondering if (yet!) another method could be added, whereby the percentile is calculated using the best-fit (I presume using moment-matching to derive k and theta) Gamma Distribution to the data?
This would allow users to very easily and simply then derive the SPI drought index using precipitation data as the standard deviation threshold for droughts maps 1-1 directly onto the relevant 32% percentile. (You can still approximate the SPI index using percentiles currently, but it will not be precisely the same due to the fitting methods used).
Thanks :-)
Hello Adrian,
Unfortunately, we don't have any statistics experts here. Do you have any idea how to implement this method?
Cheers,
Uwe
What language is cdo based on? it might be something I can try to code for integration?
CDO is written in C/C++. But a functioning implementation is helpful in any language.
Ok, I'll try to do something over Easter. 