So, what does it mean to deploy a model? Deployment is an all-encompassing term that covers the process of taking a tested and validated model from your local computer, and setting it up in a sustainable environment where it's accessible. Deployment can be handled in a myriad of ways; in this chapter, we'll focus on the knowledge and best practices that you should know about to get your models up into production.
Your choice of deployment architecture depends on a few things:
- Is your model being trained in one environment and productionalized in another?
- How many times are you expecting your model to be called predictions to be made from it?
- Is your data changing over time or is it static? Will you need to handle large inflows of data?
Each of these questions can be answered by breaking down our model selection options into two buckets...
