Thanks @hamsal !!!

At this juncture, I would like to confirm that we think sparse multioutput data is a common enough occurrence to use sparse matrices for its representation. Until #3350, I had assumed we were only interested in sparse representation of multilabel data, where spareness is common.

@jnothman Do you suggest supporting sparse multilabel data and dense multioutput multiclass, but not sparse multioutput multiclass?

I think jnothman is for supporting sparse multioutput multiclass since he suggested that sparse label encoding code from the knn PR be implemented here

That was a conditional support: we should definitely support multioutput
LabelEncoder rather than inventing that wheel in multiple places; and if we
want support for sparse multioutput, then it should be supported by
LabelEncoder too. Whether we need support for sparse multioutput data is a
matter of finding datasets that match this expectation.

On 27 August 2014 10:49, hamsal notifications@github.com wrote:

I think jnothman is for supporting sparse multioutput multiclass since
he suggested that sparse label encoding code from the knn PR be implemented
here

—
Reply to this email directly or view it on GitHub
#3592 (comment)
.

If there's any desire to complete #3483 (#3243), someone should adopt #3483
before this. I will definitely need to abandon them if this work is merged
first.

Thanks,
Michael J. Bommarito II, CEO
Bommarito Consulting, LLC
Web: http://www.bommaritollc.com
Mobile: +1 (646) 450-3387

On Tue, Aug 26, 2014 at 9:58 PM, jnothman notifications@github.com wrote:

That was a conditional support: we should definitely support multioutput
LabelEncoder rather than inventing that wheel in multiple places; and if
we
want support for sparse multioutput, then it should be supported by
LabelEncoder too. Whether we need support for sparse multioutput data is a
matter of finding datasets that match this expectation.

On 27 August 2014 10:49, hamsal notifications@github.com wrote:

I think jnothman is for supporting sparse multioutput multiclass since
he suggested that sparse label encoding code from the knn PR be
implemented
here

—
Reply to this email directly or view it on GitHub
<
https://github.com/scikit-learn/scikit-learn/pull/3592#issuecomment-53513685>

.

—
Reply to this email directly or view it on GitHub
#3592 (comment)
.

@mjbommar I can finish your work, I will do it here #3599

Thanks!

Thanks,
Michael J. Bommarito II, CEO
Bommarito Consulting, LLC
Web: http://www.bommaritollc.com
Mobile: +1 (646) 450-3387

On Thu, Aug 28, 2014 at 1:11 PM, hamsal notifications@github.com wrote:

@mjbommar https://github.com/mjbommar I can finish your work, I will do
it here #3599 #3599

—
Reply to this email directly or view it on GitHub
#3592 (comment)
.

This is implemented in OneHotEncoder afaik. If not, please open an issue describing the use-case.