I have a dataset with 15 input columns, and 4 output column names. These 4 outputs are related to each other.

I have fitted four different GLM models to predict them.

But I want neural networks/random forests/other machine learning techniques.

What would be the best approach for this problem if I want to have:

  1. A single model with multiple outputs (Continuous values)
  2. Preferably gives me the distribution of the output rather than a single value.
  3. Takes into account the correlation in the output space. These four values are somewhat related to each other. X1>X2>X3>X4 always.

Any articles/guides etc would be greatly appreciated.

  • $\begingroup$ Neural networks should work fine. As well as multivariate logistic regression (multiple dependent variables). $\endgroup$ – Jon Dec 7 '16 at 23:45
  • $\begingroup$ Would love some more details.... $\endgroup$ – maximusdooku Dec 8 '16 at 5:00
  • 1
    $\begingroup$ @maximusdooku What type of outputs are these? Continuous value, binary? And how do the four outputs relate to each other? $\endgroup$ – horaceT Dec 8 '16 at 6:25
  • $\begingroup$ @horaceT Continuous values. I have provided more info in the question. But these are basically levels of flood. $\endgroup$ – maximusdooku Dec 8 '16 at 18:37
  • $\begingroup$ Would love to get any responses. $\endgroup$ – maximusdooku Dec 12 '16 at 21:00

Your Answer

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

Browse other questions tagged or ask your own question.