Predicting customer demand to help manage inventory and resources appropriately is a challenging problem. In this project, we use data from pickups made by Uber in New York City to create a model to predict demand given the time and location.

There will be three parts to this demo:

1. Analysis and publishing
2. Scheduling
3. Deploying the model as an API

Analysis and publishing

In this part of the session we'll perform an analysis and then publish these findings as an easily-consumable Report for collaborators and business users.

Procedure:

1. Import modules for our analysis. Some of these come pre-installed in the DataScience Cloud environment. For the rest, we will install them using install.packages().
2. Load our data.
3. Run some analyses.
4. Publish the resulting analysis as an attractive Report.

Scheduling

Next, we’ll show you how to schedule a job so that you can easily automate processes that need to happen on a regular basis (like ingestion of new training data, batch data transformations, and updated predictions).

Deploying the model as an API

Last (but definitely not least), we'll deploy a model that predicts the expected demand given a time, date, and location. By deploying the model as an API anyone in the organization with an API key can query it, including any apps that the engineering team has developed.