none
Multiple Columns of Output in Azure Machine Learning Studio RRS feed

  • Question

  • Hi,

    I am trying to using the Azure Machine Learning Studio to build the model for my project. The project is to optimize the weight of the stock in an invested portfolio based on the stocks' risk and return ratio. In the dataset, it has columns as the following:

    Portfolio number, Stock1, Risk Ratio1, Return Ratio 1, Weight 1,  Stock2, Risk Ratio2, Return Ratio 2, Weight 2,  … Stock 10, Risk Ratio 10, Return Ratio 10, Weight 10, Portfolio Return Ratio.

    I tried to used couple different algorithms, like Neural Network and Multi-Variate Regression, to build the models.  In the project, I want to obtain multiple columns of output which are the weights of all the stock in the portfolios. However, it looks like the output of all the models can only be one numeric output for one row of input data. For example, the input data is the information listed in the dataset above and the output from the model would be all the weights for the stocks in one portfolio. Are there any solutions can help with it? Or Azure Machine Learning Studio can only provide one column of output instead? 

    Thanks,

    Doris

    Friday, September 6, 2019 5:55 PM

All replies

  • Hi,

    Thank you for your feedback. Unfortunately, multi-label regression isn't supported in Azure Machine Learning at the moment. Hence, you would need to create a separate learner for each predictive outcome. Another option may be to utilize our R/Python modules to train the model using multiple dependent variables. For your reference, please feel free to review our documentation as well as trained model examples in the Azure AI Gallery. Hope this helps. Thanks.

    Regards,

    Azure CXP Community.

    Friday, September 6, 2019 10:16 PM
    Moderator