What is the difference between discrete and continuous data in Tableau?
Have you ever made a graph in Tableau with an unexpected result? This may be because you needed to convert your data type to discrete when it was continuous, or (more likely) you needed to convert it into continuous when it was discrete. This blog covers dealing with continuous and discrete data in Tableau.
Firstly, if you don’t know the difference between discrete and continuous data, you can read about this in my previous blog.
Secondly, I would like to address that many people will conflate the terms continuous and discrete with measures and dimensions in Tableau. Some people prefer to use the terms measure and dimensions and will never use continuous and discrete, while others will do the opposite and still some others will use them interchangeably.
Measures are not always continuous.
Dimensions are not always discrete.
In order to demonstrate the unexpected results you can get when your data is in the wrong form, I have created a simple data set of sales from two fictional stores.
NB: For illustration purposes and to make things more obvious, store 1 has the exact same sales pattern each year. It sells $100 worth in January, $200 in February, $300 in March etc. until December, where the pattern is the repeated the following year.
When working with time and a measure, we expect that we will see how a particular measure has changed over time. We expect to see this plotted out on a line chart. With discrete time data, however, we see an aggregation of values within that particular time part.
Coloring things by discrete will give completely separate colours, while if you use continuous, you will get a colour scale.
Measures can/will aggregate, even if discrete. However, dimensions will convert to continuous measures if you try to aggregate them (add video).
Unexpected results with time
Let me first demonstrate the difference between continuous and discrete data with some line graphs.
The following graphs all show the exact same data – total sales over time. However, as you can see, the way this data displays can be quite different. This depends on the way you have set it up – if you are allowing your data to be discrete or continuous.
Annual Granularity
The first two graphs show total sales over 3 years, and are the most similar you will see in the series. The one on the left is using time as a discrete measure (note the blue pill) whereas the one on the right uses it as a continuous measure (green pill). Note that because the sales are recorded per month, but these graphs only show the sales amount per year, the amount displayed is an aggregated amount for each year.
 |
 |
Monthly Granularity
We can then see what happens when we look at the sales for each month.
Note that the graph below has aggregated the amount per month for each year. At first glance we think we are seeing things over the whole time period, but you will note when looking at the x axis, that there is no reference to years, only months. Furthermore, the amount in January is $300, rather than $100. This is because the amount for each January across all years has been summed together to give an aggregate amount. This happens for each month because we are using a discrete monthly measure in the columns shelf.
If we want to keep the years separate, we can add a discrete yearly measure before the monthly measure and the data will show the months across all years. This is almost what we expect to see, but not quite.
Once we change the measure to a continuous monthly measure in the column shelf, then we see the classic timeline data as expected.
I hope this has given you a better insight into dealing with continuous & discrete data in Tableau.