I have a queuing model written in Scala where different categories of people end up a different queues. We have a dataset providing a map of features to the numbers of people ending up at each queue, ie multiple inputs to multiple outputs (continuous values)

I have some experience using mllib for single value predictions in Scala but I can't see that multiple outputs are supported. It doesn't even look to me that mllib has continuous value output support as I can't see how to get a layer without an activation function.

Does anyone know of an ml library that can both support multidimensional regression (please correct me if I have the terminology wrong) and integrate with Scala?

Going forward I'd also love to try an RNN as our data has a time sequence element to it, but I believe this is even rarer in ml libraries.


A regression technique that allow to predict a multidimensional output can be the PLS ( partial least square ). I implemented it in scala and it will be soon available on Clustering4Ever repo. In fact we went a bit further by applying it with the clusterwise pattern which generate k-clusters driving by PLS regression which result with one regression model per cluster which can be much more accurate but also most costly due to approximatively 1000.n regressions. You can look on it with, A new micro batch approach for partial least square clusterwise regression.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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