python - TfidfVectorizer and sublinear_tf scaling for feature extraction in Azure ML -

i working on ml document classification problem. know how n-gram tfidf feature extraction , sublinear_tf scaling in azure ml.

in past did insci-kit learn using tfidfvectorizer (see example below) problem in azureml cannot explicitly define own methods or classes using python module , rather not upload zipped code.

i python person open using r if there equivalent. there r sample in marketplace dependent on unigrams.

tfidfvectorizer(max_df=.67,min_df=.015,lowercase=false ,sublinear_tf=true,norm='l2',tokenizer=abstracttokenizer()) 

best, -ari

welcome using azureml.

for problem of defining own methods, azureml visual flow-based programming tools ml modeling. it's different programming @ localhost. can define modules different dataset process, , link them draging , droping line linked between 2 modules. there list of existing modules on ml stduio, refer https://msdn.microsoft.com/en-us/library/azure/dn906033.aspx. need combine them make own ml model.

for problem of using python module, there limits python on azureml. can't costomize python installation, refer 4th item of https://azure.microsoft.com/en-us/documentation/articles/machine-learning-execute-python-scripts/#limitations.

however, r language, can import packages not installed in ml studio. please refer https://azure.microsoft.com/en-us/documentation/articles/machine-learning-extend-your-experiment-with-r/#importing-packages.

meanwhile, can author custom r modules in azureml. please refer https://azure.microsoft.com/en-us/documentation/articles/machine-learning-extend-your-experiment-with-r/#importing-packages.

so think can run r sample installing dependency "unigrams" thur api install.packages('<pkgs.zip>', ...).

best regards.