HyukjinKwon commented on pull request #33954: URL: https://github.com/apache/spark/pull/33954#issuecomment-917742920
When the index type is not specified, the default index would be attached by default for the output DataFrame. Using default index usually brings some performance penalty. See the pseudo code below. Assume we're using distributed sequence index (https://koalas.readthedocs.io/en/latest/user_guide/options.html#default-index-type), and we perform an panda API such as: ```python df.apply(func) ``` Internally the Spark operations as below would be performed (pseudo code) **Without index type hint:** ```python spark_df = df.to_spark() # compute date part spark_df.select(udf(func, "data column types without index")) # compute index part spark.createDataFrame(spark_df.rdd.zipWithIndex().map(lambda p: p[1], p[0])) # attach index ``` **With index type hint:** ```python spark_df = df.to_spark() spark_df.select(udf(func), "column types with both data and index") ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
