A Sankey diagram is commonly used to visualize the relationships and flows between multiple elements. Being inspired by the blogs on Sankey Charts in Tableau, I made an attempt to build one using student page_views data that was gathered in a MOOC course. The diagram shows course participants content access flow and potentially suggests certain patterns.

The diagram was built with two data points that are included in a page_view object:

• user_id: a course participant that clicked on a course object(page, tab, menu, link, etc.)
• content type: the type of a content object that an user clicked on

Steps:

1. preparing the data file: user_id, content_type, RowType (‘original’ or ‘duplicates’)
2. create a new field [ToPad] based on ‘RowType’:

   if [RowType]=='original' then 1 else 49 end

3. create a new Bin of Size 1 called [Padded]
4. create a third function [t]:

   (index()-25)/4

5. build functions that will show our data at the right points vertically when we build the Sankey, these are identical:

   [Rank 1] = RUNNING_SUM(COUNTD(user_id))/TOTAL(COUNTD(user_id))

   [Rank 2] = RUNNING_SUM(COUNTD(user_id))/TOTAL(COUNTD(user_id))

6. start with a sigmoid function – the basis of the Viz (that gives the curve) [sigmoid]:

   1/(1+EXP(1)^-[t])

7. create the curve [Curve]:

   [Rank 1]+(([Rank 2] - [Rank 1])*[Sigmoid])

Resources: http://www.theinformationlab.co.uk/2015/03/04/sankey-charts-in-tableau/