All right, now we finally have everything set up, so we can have a look at the data. I’m going to use SQL to read it. Copy and paste this SQL code. Notice that it starts with %sql so Spark knows this is SQL instead of Python.

This code reads data from the radio.json file we uploaded earlier into a table called radio. First, it checks if a table called radio already exists, and if it does, then it drops the table. Of course, we know that it doesn’t exist because we haven’t created it yet, but if we were to run this cell a second time, then it would exist, so it’s always a good idea to start by dropping the table. Notice that we’re able to use this simple pathname here because we mounted the filesystem in the last lesson. If we hadn’t done that, we’d have to use the long URL that starts with abfss.

Next, we’ll do a SELECT * to see all of the data. This is a really small dataset, so there aren’t very many rows to show. These records show details about a sample of users of an online radio app. The level column shows whether they have a paid subscription or not. There’s also a location column, so let’s see which locations have the most paid subscribers.

Click the chart button down here. This isn’t a very useful chart, so click Plot Options. First, remove the columns that it selected. Then drag location to the Keys box, level to the Series groupings box, and level again to the Values box. Then change Aggregation to COUNT. That looks better. Now click Apply.

Okay, have a look. Do you notice anything weird? It might not be obvious at first, but none of the locations have both free and paid users. Let’s look at the data in a different way. Go into Plot Options again and change the Display type to Pie chart. There are only four locations with paid users, and none of those locations are in the free users pie chart.

Okay, that was a pretty simple analysis, but now you know how you can work with ADLS data from Azure Databricks. And that’s it for this lesson.