This is part two of the blog about creating a blend and scaffold with Tableau Desktop, for the first part, click here.
Making a scaffold can be done quite quickly in most cases. Back to the available material: there are two sources, one with sales data and another with customer contact information.
Making a blend
The overlap between both sources is in time (the column with dates) and location (the city where the shop is located). In principle, we create a calendar file with all dates in the first column, but because we want to know what happened in each shop, that value or category must also appear in the calendar file.
If Tableau does not automatically link the date columns, you can do it manually by clicking on "Data" => "Blend relationships" (see below). Select the primary source in the drop-down menu and the secondary source under 'Secondary data source:'. Link fields can be added with 'Add...'.
NB Only dimensions can be linked in a blend.
Next, the date fields that occur can be manually linked so that the primary and secondary source can be linked at any level (see below). In the data pane, the dimensions are now followed by red chains (connected) or grey chains (disconnected). The blue circle next to a source indicates that it is a primary source. An orange circle indicates a secondary source.
If we now create a simple table like the one below with both sources having date and location/shop linked, the end result looks like this:
PS Why has a filter on 'Date' been added? Actually, because the calendar file contains the entire month of August and September, while there are only dates from 8 August. This would distort the picture somewhat.
Why do we make these choices in a blend?
Why do both time and location have to be in the scaffold/calendar file? This can be shown in Tableau quite quickly by clicking the symbol for 'link' (a chain), this symbol looks like this:
First, we set the link to 'date' in the secondary source customer contact:
The dimension/category date or time is ignored and all messages per 'Week or Date' are added together. Only locations or shops are distinguished.
The following happens when turning off or unleashing 'location':
Here, the shop is considered, but all complaints per week are counted together.
In short, this is the way to make a blend with a scaffold without additional programmes.