You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently SAM supports PMML as a model exchange format.
PMML is very limited and the extension of the format takes months/years.
PFA (Portable Format for Analytics) describes and exchanges predictive models produced by analytics and machine learning algorithms.
PFA is not limited and any new model can be defined.
PFA is supported by Spark, Python (Titus), Knime (currently working on it), etc.
As far as I know there is also Storm support for PFA - but I'm not sure about that.
It becomes especially interesting when SAM has runners for Spark (announced at the DataWorks Summit 2018 in Berlin), because all algorithms of the MLLib can be addressed by PFA, which is not possible with PMML.
The text was updated successfully, but these errors were encountered:
Currently SAM supports PMML as a model exchange format.
PMML is very limited and the extension of the format takes months/years.
PFA (Portable Format for Analytics) describes and exchanges predictive models produced by analytics and machine learning algorithms.
PFA is not limited and any new model can be defined.
PFA is supported by Spark, Python (Titus), Knime (currently working on it), etc.
As far as I know there is also Storm support for PFA - but I'm not sure about that.
It becomes especially interesting when SAM has runners for Spark (announced at the DataWorks Summit 2018 in Berlin), because all algorithms of the MLLib can be addressed by PFA, which is not possible with PMML.
The text was updated successfully, but these errors were encountered: