1
$\begingroup$

Consider a problem like this You have a customer profiling application(say classic telecoms data) You have customer transactions(lots of them) you want to find rules There is a data element which is sparsely populated ex Pets you are interested in it because you want to explore a hypothesis 'Are cat owners wealthier than dog owners?'

So you have a lot of data.But the data element you are interested in (pets) is sparse

and you have to create training data for this situation ..

How would you go about doing it?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

Consider to use some "sparse" statistical tools like sparse principal component analysis, sparse linear discriminant analysis etc.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.