Jeff Zhang created SPARK-12942:
----------------------------------
Summary: Provide option to allow control the precision of
numerical type for DataFrameWriter
Key: SPARK-12942
URL: https://issues.apache.org/jira/browse/SPARK-12942
Project: Spark
Issue Type: Improvement
Components: ML, SQL
Reporter: Jeff Zhang
This is useful for libsvm where most of features are double, and this can
reduce the file size and gain some performance improvement when reading the
libsvm file in training.
So it can be libsvm specific or global setting which apply to other data
sources. We can discuss it here. \cc [~mengxr]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
