AutoViML / Auto_ViML

Automatically Build Multiple ML Models with a Single Line of Code. Created by Ram Seshadri. Collaborators Welcome. Permission Granted upon Request.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

running Auto_ViML with python interpreter throws an ipython exception.

manugarri opened this issue · comments

Im testing Auto_Viml as one of automl containers in my pipeline. However, running Auto_Viml function from a script (or importing it from standard python REPL) throws the following exception:

    from autoviml.Auto_ViML import Auto_ViML
  File "/usr/local/lib/python3.7/site-packages/autoviml/Auto_ViML.py", line 29, in <module>
    get_ipython().magic(u'matplotlib inline')
NameError: name 'get_ipython' is not defined

This can be fixed by either using a jupyter notebook (which i imagine is the only thing tested so far) or using ipython instead of python. This is ok for toy examples but in production systems python is the default executable.

Would make sense to make ipython magic not fail.

I am using autoviml==0.1.651

Hi:
Thanks for trying out Auto_ViML. I have not been too keen on removing get_ipython since a lot of the benefits of Auto_ViML are for trying out different models and techniques to find what works best on your data set. This kind of experimentation works well only a Jupyter Notebook. Hence I don't find the need to remove get_ipython yet. I might make a version available once I get the pipeline working. Until then, this is going to be strictly for those data scientists who want to build a challenger model using Jupyter Notebooks. Hope this clears any misunderstandings.
thanks

Got it, i think i will stop using the (very promising) project until then, in my case there is no point in testing models that i cant use afterwards.