Explain Scatter Plots, Aggregation, and Granularity?

[vatsallya-lamgadey-926c32c9 (Vatsallya Lamgadey)]

If you place one measure on the Rows shelf and another measure on the Columns shelf, you are asking Tableau to compare two numerical values. Typically, Tableau chooses a scatter plot as the default visualization in such cases. The initial view will most likely be single mark, showing the sum for all values for the two measures. This is because you need to increase the level of detail in the view.

Start building the scatter plot

There are various ways to add detail to a basic scatter plot: you can use dimensions to add detail, you can add additional measures and/or dimensions to the Rows and Columns shelves to create multiple one-mark scatter plots in the view, or you can disaggregate the data. And, you can also use any combination of these options. This topic looks at these alternatives using the Sample-Superstore data source.

To create the initial view, follow these steps:

1. Place the Sales measure on the Columns shelf.
2. Place the Profit measure on the Rows shelf.

The measures are automatically aggregated as sums. The default aggregation (SUM) is indicated in the field names. The values shown in the tooltip show the sum of sales and profit values across every row in the data source.

Follow the steps below to use dimensions to add detail to the view and to disaggregate data.

Use dimensions to add detail

Follow these steps to develop the scatter plot view you created above by adding dimensions to show additional levels of detail.

1. Drag the Category dimension to Color on the Marks card.This separates the data into three marks—one for each dimension member—and encodes the marks using color.
   
2. Drag the State dimension to Detail on the Marks card. Now there are many more marks in the view. The number of marks is equal to the number of distinct states in the data source multiplied by the number of categories.
   

Although more marks are now displayed, the measures are still aggregated. So regardless of whether there is one row in the data source where State = North Dakota and Category= Furniture, or 100 such rows, the result is always a single mark.

Maybe this process is developing the view in a direction you find useful, or maybe you prefer to go in a different direction—for example, by adding a time dimension to the view, or by introducing trend lines or forecasting. You decide what questions to ask.

Try adding more fields to the rows and columns shelves

Revert to the original one-mark view and follow these steps to develop the scatter plot view by adding fields to the Rows and Columns shelves.

1. Drag the State dimension to the Columns shelf.Even if you drop Continent to the right of SUM(Sales), Tableau moves it to the left of SUM(Sales). This is because you cannot insert a dimension within a continuous axis. Instead, your view shows a separate axis for each member of the dimension.
   
2. Drag the Segment dimension to the Rows shelf. You now have a view that provides an overview of Sales and Profit across states and customer segments. It can be interesting to hover over the marks in the view to see tooltip data for various segments:
   

Try disaggregating the data

Another way to modify your original one-mark scatter plot to display more marks is by disaggregating the data.

Clear the Analysis >Aggregate Measures option. If it is already selected, click Aggregate Measures once to deselect it.

What you have actual done is to dis-aggregate the data, because this command is a toggle that was originally selected (check mark present). Tableau aggregates data in your view by default.

Now you see a lot of marks—one for each row in your original data source:

When you disaggregate measures, you no longer are looking at the average or sum for the values in the rows in the data source. Instead, the view shows a mark for every row in the data source. Disaggregating data is a way to look at the entire surface area of the data. It’s a quick way to understand the shape of your data and to identify outliers. In this case, the disaggregated data shows that for many rows in the data, there is a consistent relationship between sales income and profit—this is indicated by the line of marks aligned at a forty-five degree angle.