How to eval spark.ml model without DataFrames/SparkContext? -
with spark mllib, i'd build model (like randomforest), , possible eval outside of spark loading model , using predict on passing vector of features.
it seems spark ml, predict called transform , acts on dataframe.
is there way build dataframe outside of spark since seems 1 needs sparkcontext build dataframe?
am missing something?
re: is there way build dataframe outside of spark?
it not possible. dataframes live inside sqlcontext living in sparkcontext. perhaps work around somehow, whole story connection between dataframes , sparkcontext design.
Comments
Post a Comment