search

JeffreyRacine / R-Package-np

R package np (Nonparametric Kernel Smoothing Methods for Mixed Data Types)
https://socialsciences.mcmaster.ca/people/racinej
47 stars 18 forks source link

Feature Request #35

Closed Jadamso closed 5 months ago

Jadamso commented 2 years ago

I have some requests to improve an already great package.

To eliminate some dependency on stats::loess and other packages, npreg could take other common kernels: "tricubic", "triangular", "indicator/histogram".

JeffreyRacine commented 5 months ago

Greetings,

Just to let you know that "indicator/histogram" kernel is available via "uniform"...

Jadamso commented 5 months ago

I was a bit unclear and should have specified these as two separate requests:

  1. ckertype='tricubic' kernel. I want to replace any dependency on stats::loess.
  2. ckertype= 'histogram', which uses semi-open intervals '[ , )' and sets corresponding design points. I want to use your infrastructure for bandwidth selection in histograms and regressograms.

Your package is great and these features will reduce some initial barriers to using it, in my opinion. But I also understand why you might not want to add such features, which is why I closed the issue.