If you have a dense time-series data without seasonality, one of
the most effective visualization techniques is the horizon chart. This advanced chart does require some complex
calculations to create. Thankfully, Marc
Reid has written a blog post on what the horizon chart is about and how to
create it.
Marc also has published a video to show how to copy over the calculations and worksheets
to your own workbook to create your horizon chart. I followed the steps in his video and
created the above horizon chart rather easily.
Note that Marc Reid’s post referred to
Yvan Fornes’s blog post in which Yvan wrote about setting the stack marks
option to Analysis > Stack Marks > On.
I think this is a misstatement as you want to set the stack marks to ‘Off’
(not ‘On’).
Tableau describes stack marks as ‘When marks are stacked,
they are drawn cumulatively along an axis.
When marks are not stacked, they are drawn independently along an axis.
That is, they are overlapping.’ That’s
what you want in the horizon chart where the marks in area chart should
overlap. So set the stack marks to
‘Off’.
Now let's compare between cycle plot (for time-series data with seasonality) and horizon plot (for dense time-series
data without seasonality) to see why different types of data require different types
of visualization.
If we were to visualize the bond yield data for 18 countries (which is a
dense time-series data without seasonality) as a cycle plot, we would get something
like this.
This graph is
considered not useful because it would take too much time for the users to
determine useful trends. So the job of a
data visualization designer is to know and apply the appropriate type of
visualization for the specific type of data.
No comments:
Post a Comment