In our case we will create an API with just one endpoint. For example, this Football API ( link) for example, offers us info about seasons at this endpoint ( link), but also team info at this other endpoint ( link).Įach endpoint has a different URL and it might (or not) require different parameters. In our case, our algorithm will receive data in order to predict, so we will use a POST method.īesides, an API can have several endpoints, that is, different ways by which we can connect with the other service. POST: it allows us to create or send data.Now, what types of actions or methods can be done? Well, there are mainly the following:
If the order has been sent correctly, this second service executes an action and returns a result. If you saw the webinar on how to work with APIs in R or, you already know about APIs, you can skip this section.Īs we have said before, an API allows two computing services to communicate. It sounds cool, right? Sure it does! So, let’s learn how we can put our R models in production! How to build an API with Plumber Understanding the basic concepts of an APIīefore we go in and see how to create an API, it is important to first understand some fundamental concepts of an API. The idea is simple: when they submit the form correctly, they are told which type of flower it is. The idea is that users will have a form on a website with those same variables as placeholders. Imagine that we have trained an algorithm that classifies flowers depending on four variables. As I explained in this webinar at Deusto University (sorry it is in Spanish), an API is a way that enables two computing services to communicate. But sometimes our R codes are not auto sufficient: some other services might need to use them. In this blog, we have already learned some ways of squeezing out our R scripts, from automating them in Google Cloud or doing so in Windows or Mac.
To create a machine learning algorithm is cool but… what’s next? How can I put my model (in this case made in R) in production?
0 kommentar(er)
