Re: mahout spark-rowsimilarity error

[Pat Ferrel]

Some ideas are left on the SO question

On May 18, 2017, at 6:14 PM, Daniel Gabrieli <dgab...@salesforce.com> wrote:

Hi,

I am trying to research potential features to feed into the Universal Recommendation engine by using the mahout command line interface.  I can successfully run mahout spark-rowsimilarity on up to about 500 rows of data. After that I get the error below:

INFO DAGScheduler: Job 7 failed: saveAsTextFile at TextDelimitedReaderWriter.scala:294, took 1.110184 s

Exception in thread "main" org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 12.0 failed 1 times, most recent failure: Lost task 0.0 in stage 12.0 (TID 24, localhost): java.lang.IllegalArgumentException

I tried various cli options like below but get the same error:

mahout spark-rowsimilarity --maxObservations 500000 -sem 6g -ma "local[4]" --input items.csv --output /tmp/output

Any suggestions would be most helpful.

I cross posted on Stack Overflow here

Thanks

[Pat Ferrel]

Using the Mahout CLI is not a typical approach to research features though it might be useful. 

What I always ask is: what is the feature, what does it indicate about the user, how many possible items for the feature type, how many users have indicated this feature.

average users/item and average items/user for the feature are also useful when compared to the same for conversion or primary feature/indicator.

--
You received this message because you are subscribed to the Google Groups "actionml-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to actionml-use...@googlegroups.com.
To post to this group, send email to action...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/actionml-user/1C2E66DF-BD3B-4584-B30B-5934D0E90B84%40occamsmachete.com.
For more options, visit https://groups.google.com/d/optout.