Skip to Content

Multivariate Time series Modeling in SAP PA Automated Analytics

Team

Is it possible to do multivariate time series using automated analytics in SAP PA.

I did try modeling using multiple variables it seems like it only considered one extra variable rest were ignored.

We use VAR models in R to generally model these type of scenarios.

Even in Expert Analytics for time series only Triple exponential smoothing is available , I am not sure we can model it here as well.

As a solution for this we will have to use R algorithms ??

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

2 Answers

  • Best Answer
    Feb 08, 2017 at 01:58 PM

    Hi Sainath, the short answer is you cannot have multiple target in Automated Analytics time series analysis.

    See (and vote?) the idea here:

    https://ideas.sap.com/D39099

    Kind regards

    Antoine

    Add comment
    10|10000 characters needed characters exceeded

  • Feb 08, 2017 at 04:41 PM

    Hi Antoine

    Let me clarify on what I was asking with an example. I have time series data of the variable that I need to predict (which in this case would be the target variable) . Most of the time this data itself (time based lag data of this variable) is used to predict the and build the model . I believe this is called Uni-variate Time series Analysis.

    In my case I have predictor variables with time series data as well , in this case I have multiple variables to predict the target variable .

    Therefore when I use the time series based predictor variables to predict the target the automated analytics is automatically using only the default functionality . In R I would use VaR -Vector Auto regressive models to do such type of modeling.

    Add comment
    10|10000 characters needed characters exceeded