Are there plans for a go aka golang wrappers or package?

[arnwas]

Are there plans for a go aka golang wrappers or package?

So similarly as those for python or R?

Greetings, Arno

Summary

Motivation

Description

References

[StrikerRUS]

@arnwas Hi! As I know, we don't have any plans for full Go package.

However, there is a Go package for prediction part with LightGBM models. Will leaves package be enough for your task?

Can you please describe your use case?

[arnwas]

My use case would simply include the training part, which is not provided by leaves unfortunately.

[StrikerRUS]

Ah, I see!

Please elaborate why you want to use Go. It will allow to prioritize contributors' efforts. Otherwise, we may fall into a situation of the need to implement wrappers for all existing programming languages.

[arnwas]

I am already using Go, mainly for speed reasons, when big amounts are to be prepared, and python's friends e. g. numpy are too slow and consume too much memory. On the other hand i realized lightGBM giving better models, so i wondered how i would combine those two. Rewriting my things in python is not really an option, as i had to give up python for above reasons. If you have suggestions for workarounds feel free to tell me!

[StrikerRUS]

Thanks for the explanation!

Seems that dmitryikh/leaves#64 can be treated as upstream issue. However, I'm adding this request in our hub.

[StrikerRUS]

Closed in favor of being in #2302. We decided to keep all feature requests in one place.

Welcome to contribute this feature! Please re-open this issue (or post a comment if you are not a topic starter) if you are actively working on implementing this feature.

[mirekphd]

Rewriting my things in python is not really an option,

@arnwas : out of genuine interest: would someone interested in following your lead and switching to Go be able to find pandas, numpy, scipy, sklearn, (mlflow, redis, minio) equivalents in Go?

[arnwas]

He would, although those might be less well-known. But not all projects do need those, as mine does hardly at all. The only reaason to use numpy would - in my case - be sheer performance, an go does it by itself as opposed to python.