-
Notifications
You must be signed in to change notification settings - Fork 191
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
transform(t::Tuple, x) #781
Comments
Could you describe the use case? Currently, supported transformations would cancel each other if applied sequentially. |
I want to use it for centering and normalizing before applying a PCA. The PCA from MultivariateStatistics.jl only supports centering but not standardizing and they don't want to support it (JuliaStats/MultivariateStats.jl#86). I really need that feature and don't want to pirate the methods (https://github.com/gdkrmr/WeightedOnlineStats.jl/blob/master/src/pca.jl#L58). MultivariateStats.jl is going to use |
Ah so PCA would be the second There's also an API question about what to return. If you want to be able to Previous issue at #452. |
Yes
I am bit lost here, so the |
Yes currently |
Aren't they planning to change that? |
I'm not sure. Why don't you ask them? |
|
Agreed, there are some open API questions. You don't need the intermediate steps to reconstruct the input unless you want to support |
Well @wildart We should probably define |
Yes, we should definitely put |
I think such an abstraction could be quite handy, just apply all the data transforms one after another to x. The same should exist for
reconstruct
. Would you be interested in such an extension? I could try to implement it.The text was updated successfully, but these errors were encountered: