Skip to content

Deployment of Machine Learning Algorithms on Microsoft Azure ML studio

Notifications You must be signed in to change notification settings

vishalsatam/DeploymentOfMLAlgoOnCloud

Repository files navigation

Deploying Machine Learning Services on Cloud

Assignment 3

Team 1 - Manasi Dalvi & Vishal Satam

This project has been created to deploy the classification and prediction algorithms that we have developed for the Fredie Mac's dataset. More details available on the github URL : https://github.com/vishalsatam/PredictiveModellingOnFreddieMacLoans You will require login credentials from Freddie Mac's Single Family Loans dataset http://www.freddiemac.com/research/datasets/sf_loanlevel_dataset.html in order to execute the below docker image.

Docker Image

The docker image has been created for preprocessing the data from Freddie Mac's website.

Pull the image

docker pull vishalsatam1988/assignment3

Run the summarization script

docker run -it vishalsatam1988/assignment3 sh /src/assignment3/downloadAndClean.sh "<username>" "<password>" <startyear> <endyear>
Eg : docker run -it vishalsatam1988/assignment3 sh /src/assignment3/downloadAndClean.sh "[email protected]" "Eq=yF?f3" 2005 2016

Commit the running container

docker commit <containerid> vishalsatam1988/assignment3
  • You can use the origination file to upload directly to Microsoft Azure for building the Predictive Models.
  • For the performancesummary.csv file located at /src/assignment3/data/ you would have to do some pre-processing to perform Random Undersampling on the majority class before using the file in Microsoft Azure.
  • For this process, additional memory is required and your docker machine might return a Segmentation Fault. So, you can run these pre-processing functions from the jupyter notebook.

For pre-processing the performancesummary.csv file, please open the Jupyter notebook to access the functions for pre-processing and use the function create_train_test_sample() to create the train and test files that can be uploaded to Microsoft Azure for building the Classification Models.

View results in Jupyter Notebook - Open /src/assignment3/RandomUnderSampling.ipynb

docker run -it -d -p 8888:8888 vishalsatam1988/assignment3 /bin/bash -c 'jupyter notebook --no-browser --allow-root --ip=* --NotebookApp.password="$PASSWD" "$@"'

Instructions for building Web application have been given in the Flask Application folder

About

Deployment of Machine Learning Algorithms on Microsoft Azure ML studio

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published