Focusing first on infrastructure and monitoring
Successfully applied ML projects depend on an iterative approach to tackle data collection, data cleansing, feature engineering, and modeling. After a successful deployment and rollout, you should go back to the beginning, keep an eye on your metrics, and collect more data. By now, it should be clear that you will definitely repeat some of your development and deployment steps during your project.
Getting the infrastructure around your ML project right will save you a lot of trouble. The key to successful infrastructure is automation and versioning, as we discussed in the previous chapter. So, I recommend that you take a few extra days to set up your infrastructure automation and register your datasets, models, and environments—all within Azure Machine Learning.
The same is true for monitoring. In order to make educated decisions about whether your model is working as intended, whether the training data is still accurate...
