Hi,
The current CppLowess does not allow weight on each data point so that more important/accurate data are weighted heavier, or allow fitting multi-variables. Additionally, it does not implement standard error estimation of the fitted values for statistical purposes. Could you please add these features? These would be used in genomic and biomedical researches:
https://github.com/yjiangnan/rosely (it currently depends on pyloess, but pyloess has several problems, e.g., the weights parameter does not affect anything. It is an ancient Fortran code so that debugging is almost impossible.
My own funding is running out and I will be soon forced out of my current position. But I think these features would really be useful. Thanks!
Hi,
The current CppLowess does not allow weight on each data point so that more important/accurate data are weighted heavier, or allow fitting multi-variables. Additionally, it does not implement standard error estimation of the fitted values for statistical purposes. Could you please add these features? These would be used in genomic and biomedical researches:
https://github.com/yjiangnan/rosely (it currently depends on pyloess, but pyloess has several problems, e.g., the
weightsparameter does not affect anything. It is an ancient Fortran code so that debugging is almost impossible.My own funding is running out and I will be soon forced out of my current position. But I think these features would really be useful. Thanks!