Rather than use a fancy purpose-built class, I was thinking that you could rather generate a series of label vectors, vector A is 1 when class a is positive and 0 when any other is, vector B is 1 when class b is positive and 0 when any other is, etc.
I don't know anything about streaming in particular so I don't know if this introduces any lag or concurrency problems, but you could perform the logistic regression on each of these label vectors independently using the classifier algorithm of your choice and then, concatenating the predictions into a dataframe, take a rowmax to do your multiclass evaluation. On Fri, Jan 19, 2018 at 11:29 AM, Sundeep Kumar Mehta <sunnyjai...@gmail.com > wrote: > Thanks a lot Patrick, I do see a class OneVsRest classifier which only > takes classifier instance of ml package and not mlib package, do you see > any alternative for using OneVsRest with StreamingLogisticRegressionWithSGD > ? > > Regards > Sundeep > > On Thu, Jan 18, 2018 at 8:18 PM, Patrick McCarthy <pmccar...@dstillery.com > > wrote: > >> As a hack, you could perform a number of 1 vs. all classifiers and then >> post-hoc select among the highest prediction probability to assign class. >> >> On Thu, Jan 18, 2018 at 12:17 AM, Sundeep Kumar Mehta < >> sunnyjai...@gmail.com> wrote: >> >>> Hi, >>> >>> I was looking for Logistic Regression with Multi Class classifier on >>> Streaming data do we have any alternative options or library/github prj. >>> >>> As StreamingLogisticRegressionWithSGD only supports binary >>> classification >>> >>> Regards >>> Sundeep >>> >> >> >
