Summary
In this chapter, we took a look at a few things from a high level—data, automation, change, infrastructure, monitoring, and rollouts. I hope that our coverage of these topics made sense to you after reading through experimentation, feature engineering, training, optimization, and deployment in the earlier chapters.
It's important to understand that your data will control and influence everything, and hence making data a first-class citizen in your company is a first great step. Hiring a VP of Data and defining standards on data quality, lineage, and discoverability are just a few of the measures you can take.
Automated Machine Learning will run the world in a couple of years. The idea is quite simple: a trained meta-model will always be better at proposing, training, optimizing, and stacking models for higher predictive performance than humans. This makes total sense; it's just another parameter optimization step that also includes the model architecture...
