AI & Analytics Engine Tutorial 4/4: Deployment

This is article 4/4 in the AI & Analytics Engine tutorial series. You are going to learn how to deploy our application so that it may be accessible to an end-user. 

Great! Look how far you have come! In the Engine tutorial so far you have learned about:

1. Creating a Project
2. Preparing our Data
3. Building an AI Application

To knit the tutorial series together and begin to derive value from our AI application, it's essential we deploy our application into a production environment.

In the scenario we've been using throughout this tutorial, this means others can receive a prediction regarding how they would have faired onboard the Titanic - once they've fleshed out the necessary details the application would require to make a prediction.

What is Deployment?

Deployment in a software context is all the activities that make a software system available to be used. For our application, we would require an endpoint that would invoke our prediction API whenever it's been called upon - the prediction API is simply our machine learning model that we have trained, evaluated, and deployed into a live environment. 

Step 1: Creating a Deployment 

We have 2 options to create a new deployment within the Engine: 

1. Using the dashboard, or the floating-action button from any of the model's details page 
2. Hovering over the evaluated models' rating on the models' comparison page

Whichever method you decide to proceed with, once you've clicked on "NEW DEPLOYMENT", a two-step menu will open up. 

In the Gif above you can see that we are given multiple options to select from; For the sake of simplicity, we will advance with the PI.EXCHANGE Cloud environment.

Step 2: Selecting an Endpoint

In tech-speak, an endpoint describes a remote computing device that allows back and forth communication between that device and the network to which it is connected.

The following steps are quite straightforward. You'll be asked whether you wish to create a new endpoint if you're using an existing one. Since this is our first deployment, we do not have any previous endpoints available to us. Therefore, we tell the Engine that we'd like to create a new endpoint then select "DEPLOY" to deploy it.

Upon deployment, you'll be given a full overview of your deployed model shown on the deployment summary page. The page supplies some basic information on the deployed model, sample code for calling the API in multiple programming languages, and also an API test.

The method of online prediction is via the API Test. To use the API test, you paste CSV data in the left column (without header and target column) and click on the "CALL API" button and observe the sample output.

Step 3: Predicting with your Model 

Wrap Up 

Over the course of these tutorials, we have covered the end-to-end process of how to create and deploy an AI application using the AI & Analytics engine. The AI & Analytics engine makes developing AI applications accessible and easy for people without a data science background, or if you're like me, and don't want to be writing code all of the time. See how you can get started with no code machine learning with this article! 

For a recap of the previous tutorials:

Part 1: Creating your first ML project

Part 2: Data Preparation

Part 3: Building an AI Application

Be sure to book a demo to find out how the team at PI.EXCHANGE can help you get your project over the line.