1. Groupings, calculations, and summary statistics
In this video, we're going to build on row-level calculations and explore how to group data, create higher level aggregations, and summaries on our entire dataset.
2. Understanding customer call escalations
Grouping data into higher levels helps us answer questions Oakmark's leadership would be interested in like "How many calls have we ever received?"", "What are the leading reasons for escalations?"" as well as trends like "Are escalation rates going up or down?""
Oakmark's call log data captures valuable details at the row level. By grouping it in Sigma, we can roll it up into summaries that highlight patterns and support decision-making.
3. Setting up table groupings
When grouping a table in Sigma, you'll choose columns on which to Group, and then have the option to do calculations based on that grouping, such as sums, averages, mins/maxes, and others.
You may need to have multiple groupings in a single table, such as grouping calls by Year, Quarter, and Month, or by Call Type and Channel.
4. Calculations on grouped tables
Sigma can also generate calculations on your grouped data. For example - calculating the count of calls received each quarter, or finding the average call duration by call type.
Under the hood, these calculations are aggregate calculations like we covered in the previous lesson, since they are operating on an entire column's worth of data.
5. Table-level summary calculations
Sometimes it's helpful to have table-level statistics such as the grand total, minimum, or maximum value of a given column. Perhaps you want to see at a glance what the minimum and maximum dates are in a call logs table, so you always know the extent of the data's history.
A good way to capture this information is with a Sigma's table summaries.
Sigma houses table summaries at the bottom of each table element. By default, a row and column count are shown, but you can add your own calculations.
6. Let's practice!
Now let's look at how to build and add calculations at multiple levels in a single table.