Skip to content
This repository has been archived by the owner on May 21, 2022. It is now read-only.

Split package into multiple and make this a Meta package #29

Open
3 of 5 tasks
Evizero opened this issue Mar 29, 2017 · 2 comments
Open
3 of 5 tasks

Split package into multiple and make this a Meta package #29

Evizero opened this issue Mar 29, 2017 · 2 comments

Comments

@Evizero
Copy link
Member

Evizero commented Mar 29, 2017

This package has come a long way, but at the moment it feels a bit like the package equivalent of a blob design. I am very happy with MLLabelUtils because it is a single concern kinda package, which makes it easier (and more fun) to design and flesh out.

I think it would make sense to make this package a convenience collection package that reexports data related ML packages.

  • MLDataPattern.jl (done). Only deals with the data access pattern, which are currently the dominating part of this package.

  • MLDatasets.jl. Self explanatory.

  • MLLabelUtils.jl. All things concerning classification labels.

  • MLDataGenerators.jl (doesn't exist yet). Move artifical data generators there. Also absorb Absorb joshday/DataGenerator.jl #9

  • MLPreprocessing.jl (in progress). This aspect is a bit neglected so far.

Last but not least it could offer some glue code. For example if MLDataUtils is the meta package it could define the needed interface for DataFrame and friends, so they just work out of the box.

@Evizero
Copy link
Member Author

Evizero commented Mar 29, 2017

parts of the current documentation should probably be moved to the JuliaML website instead

@Evizero
Copy link
Member Author

Evizero commented Apr 9, 2017

As a side note, I will remove precompilation from MLDataUtils, once the code move to MLDataPattern is complete. This is because I will add a Requires.jl dependency here, so that we can have default support for all kinds of different data sources, such as DataFrame and DataTable

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant