Support Questions
Find answers, ask questions, and share your expertise

Too Many Evaluations Exceptions from ARIMA fitModel call on Spark TS Library

New Contributor

I am getting an exception of math3.exception.TooManyEvaluationsException from the spark ts (time series) library during a fitModel call on the ARIMA object; and I am looking for some guidance on why this may be.

 

fitModel is called with (1, 0, 1, ts) and the ts vector is (28.0,1.0,1.0,27.0,59.0,145.0,93.0,121.0,134.0,112.0,146.0,143.0,182.0,105.0)

 

I know that is not a lot of samples but it seems odd that I would get an exception when trying to run on such a small vector.

 

The stack trace says it is coming out of this area of the sparkts module:
com.cloudera.sparkts.models.ARIMA$.fitWithCSSCGD(ARIMA.scala:196)

 

Is there something I should check for or another method I could use to produce a forecasted series?

 

Thanks.

 

2 REPLIES 2

Master Collaborator

(I think that practically speaking you'll get a better answer at https://github.com/sryza/spark-timeseries/issues )

New Contributor

Thank you.