This sample experiments provides example of deploying different versions of python modules like xgboost, sklearn, scipy etc. on Azure ML
Azure Ml comes with default set of python packages based on the 4 python environments. The experiments provides example of how to override and use a different version of a module in Azure ML. The same approach can be used for other packages as well. To get to the correct version of package, do the following 1. install the package locally on the same environment on a local machine ie. (Anaconda 4.0 etc.) 2. pick the package directory from 'c:\\Anaconda3\\Lib\\site-packages' path and add it to the root folder of the zip file. You can also add custom modules and pkl file in the zip file. So train the model offline and deploy the model on AzureML work perfectly fine.. **Note:** There are some packages that require local compile and does not work with simple directory copy, like xgboost. It also needs the DLLs ,so they need to copied as well along with the package directory.