p2-datascience-2021 -> p2-modeldeploy-2021

Data Stream

Requirements:

• Python v3.8+
• PIP
• Azure Account with Azure ML Studio up.
• Config file from Azure ML instance overview. To be placed in project directory.

Optional

• Docker

How to use (if running locally):

1.) Run the following command:

pip install -r requirements.txt

2.) Load jupyter notebooks (should be installed in step 1) and then open upload-to-azure-to-use.ipynb.

3.) Run the first four code blocks. On the fourth block uncomment the 2nd - 3rd line and comment out the 5th - 6th line to host your model online.

Leave the fourth block as is if you want to host locally. Note that you need Docker installed to run the local version.

commonreadaility-generate-model.ipynb

Expanded to get model files and process the Guthenberg files.

The Kaggle for this file can be opened by clicking here and also includes the csv files of the processed Guthenberg data for the scraped top 100 ebooks, results for the guthenberg data and includes the required generated models.

upload-to-azure-to-use.ipynb

Upload model generated by commonreadaility-generate-model.ipynb to Azure with the option to host locally or in Azure.

You need the following files before running this notebook:

• config.json - obtained from Azure Mchine Learning portal.
• model_0.joblib - generated from commonreadaility-generate-model.ipynb. The file is available here in the output section.
• model_1.joblib - same as model_0.joblib.