In this chapter, we learned to build two similar dashboards—a static one, with no server needed and using Altair, and a dynamic one, built from an ordinary Jupyter Notebook with arbitrary code and visualization packages, using the panel package. We discussed the pros and cons of each approach and when to select one over the other.

Either way, the dashboard is a great way to communicate your data product to your colleagues and clients. Dashboards allow us to get insights into business processes and spot issues early on. In many cases, that would make a perfect deliverable. In some cases, though, you might need to create a programmatic access point for your code, for example, a machine learning algorithm for an external application (a website, mobile app, or some analyst from their Jupyter Notebook) to use.

In the next chapter, we'll do exactly that, by building our own data-serving RESTful API, similar to the one we ourselves used not too long ago, in Chapter 6, First Script – Geocoding with Web APIs, and Chapter 11, Data Cleaning and Manipulation. Building an API allows our customers to directly access our application (for example, a predictive model) and use it on their data within their environment.