ARIMAX with spark-timeseries

3

Cloudera recently added the spark-time series library to github. According to the user docs, it definitely can fit autoregressive integrated moving average (ARIMA) models, but I see no mention of ARIMAX, which takes into account explanatory variables. Does anyone with more experience with this library know whether ARIMAX is possible in the library's current implementation? Mathematically, I understand the difference between ARIMA and ARIMAX, and I know extending it myself wouldn't be terribly difficult, but I'm right now looking for an off-the-shelf solution in spark. Can anyone recommend an alternative spark implementation?

Kyle.

Posted 2016-01-20T19:03:33.203

Reputation: 1 453

Answers

1

Blast. It looks as though ARIMAX is currently a work in progress for the cloudera/spark-timeseries community. Although some work has been done on the model development side, it doesn't appear to have been merged in.

Kyle.

Posted 2016-01-20T19:03:33.203

Reputation: 1 453