[mlpack] Association rule learning with mlpack

[Ryan Curtin]

On Fri, Jan 26, 2018 at 08:36:01PM +0600, Artem Fedoskin wrote:
> Dear ML Pack developers,
> 
> My name is Artem Fedoskin. I study Data Analytics at university as a master
> student and recently we had a lecture about Frequent Itemset Problem -
> namely Apriori and Eclat algorithms. Brief search showed me that these
> algorithms are not implemented in mlpack. Would it be useful if I implement
> them? I'm pretty interested in this area and for me it would be a good dive
> into the codebase of mlpack.

Hi Artem,

Thanks for getting in touch.  I think typically association rule
learning is more suited to sparse data like bag-of-words text data.  But
mlpack doesn't have great support for text data at the moment, and none
of the existing algorithms that we have are related to subset problems
like what you proposed.

I think it could still be nice to add support for this; however, if we
were to add it, we would need to have corresponding string processing or
other data processing utilities so that it could actually be useful in
practice.

I guess, basically, the first step if you were interested in doing this
would be to propose an API for how the user would interact with the
code, and then we can see how well we can make this fit with what mlpack
already does.

Let me know what you think.

Thanks!

Ryan

-- 
Ryan Curtin    | "No... not without incident."
ryan at ratml.org |   - John Preston