In order to prepare the data for the analysis, it is possible to use the data preparation tools that ML Workbench provides.
Data are available to dowload on Kaggle: Amazon Reviews
More information on how to prepare the data in ML Workbench
The file train.py contains the Python code to apply the machine learning algorithm to the data.
The file score.py contains the Python code to test the model created in the previous file on the test dataset.
On the docs it is possible to understand better how to execute the code.
az ml env setup -n <new deployment environment name> --location <e.g. eastus2>
Add -c or --cluster switch in the following command to set up an environment in cluster mode later.
az ml account modelmanagement set -n <youracctname> -g <yourresourcegroupname>
az ml env set -n <deployment environment name> -g <existing resource group name>
az ml service create realtime -f score_iris.py --model-file model.pkl -s service_schema.json -n irisapp -r python --collect-model-data true -c aml_config\conda_dependencies.yml
parameters are:
-
-f: The scoring script file name.
-
--model-file: The model file. In this case, it's the pickled model.pkl file.
-
-s: The service schema. This was generated in a previous step by running the score_iris.py script locally.
-
-n: The app name, which must be all lowercase.
-
-r: The runtime of the model. In this case, it's a Python model. Valid runtimes are python and spark-py.
-
--collect-model-data true: This switch enables data collection.
-
-c: Path to the conda dependencies file where additional packages are specified.
Command to gain more info about the model deployed:
az ml model show -m a76c078203d9426b8647bba2a1854944
az ml manifest show -i 256f60c7-e085-4ad8-9802-3ccacbe261e7
When it finishes:
docker ps to see the container running locally on Docker
Output from the Command Prompt:
model.pkl Successfully registered model Id: dc48590341a74de8811e48922076abc3 More information: 'az ml model show -m dc48590341a74de8811e48922076abc3' Creating new driver at C:\Users\jetiba\AppData\Local\Temp\tmpzkdxth20.py score.py service_schema.json Successfully created manifest Id: e541e271-5885-40d9-9842-e4d2fcc867f6 More information: 'az ml manifest show -i e541e271-5885-40d9-9842-e4d2fcc867f6' Creating image................................................................Done. Image ID: eb03653e-3f6c-4f70-8e28-af8322a6565e More details: 'az ml image show -i eb03653e-3f6c-4f70-8e28-af8322a6565e' Usage information: 'az ml image usage -i eb03653e-3f6c-4f70-8e28-af8322a6565e' [Local mode] Running docker container. [Local mode] Pulling the image from mlcrpacr14c72735e14d.azurecr.io/textsentapp:4. This may take a few minutes, depending on your connection speed... [Local mode] Pulling.............................................. [Local mode] Waiting for container to initialize [Local mode] Done [Local mode] Service ID: textsentapp [Local mode] Usage for cmd: az ml service run realtime -i textsentapp -d "{"input_df": [{"Text": "This is a great 1st drone and I was amazed by it. I got it for my son and he loves it."}]}" [Local mode] Usage for powershell: az ml service run realtime -i textsentapp --% -d "{"input_df": [{"Text": "This is a great 1st drone and I was amazed by it. I got it for my son and he loves it."}]}" [Local mode] Additional usage information: 'az ml service usage realtime -i textsentapp'
To test the model you can try this: az ml service run realtime -i textsentapp -d "{"input_df": [{"Text": "some new feedback to test if it works!!"}]}"