WoW | 2019W03

#WorkoutWednesday2019

W03 - 2018 Rugby League Home and Away

At last, I can do a true small multiple viz. Sure, I’ve done the same chart per different dimensions, but nothing like this. This served as a good opportunity to learn more about the methodology. First stop like any good analyst or developer: Google. Then, like many others, I ended up on an Andy Kriebel’s post and got the calculations for columns and rows.

Small Multiples

When creating a small multiple you typically want things to be presented as a square.
Example: say you have 16 stores you’re plotting data for. You want these to be presented in a 4x4 grid. How can you get this? Simple: square root. √16 = 4. Let’s use this example throughout.

Column

(index()-1)%(round(sqrt(size())))

Size: counts items in viz when using the dimensions in table calculation (16 stores)

• Note: the first item in the table calc (“specific dimensions”) needs to be the parent dimensions; in other words, order them hierarchically

Round Square root: makes it a friendly integer and helps to create the “square” view (4 columns)

Index: order of your items (1st store, 2nd store, 3rd store…)

Modulo: divides order of items by the rounded square root of the item count to put in columns left to right; first item in your index/rank is assigned 0, second team assigned a 1, and so on up to the rounded square root of the item count (in this case, it’s N%4, so 0, 1, 2, 3 repeating)

Row

int((index()-1)/(round(sqrt(size()))))

Size: (16 stores)

Round Square root: (4 rows)

Index: (1st store, 2nd store, 3rd store…)

Int: divide order of item (minus 1) by rounded square root of the item count and truncate (0/4 = 0, 1/3 = 0, etc. so they’re on the same row. Once index is a multiple of the rounded square root of the item count, you get a new row; 4/4 = 1, 5/4 = 1, 8/4 = 2, 9/4 = 2, etc.)

If desired, you could decide to sort Stores based on sales or the like to get desired order

Here the Stores are sorted in alphabetical order and Columns/Rows are computed using Stores.

Using this methodology you can accomplish this week’s Workout Wednesday. The lollipop charts are created as expected: circle and bar charts with synchronized dual axis. These are separated by Measure Names/Values: running totals of wins, draws, and losses.

The trickier part is the team label and stats which utilizes a methodology I often find myself using: AVG(1). In this case, we change the mark type to a line so it stretches across the viz. Drag all the info you want to display onto the text label and check:

• Allow labels to overlap other marks

• Label start of line (and change text alignment to “Left” so it’s as far left as possible)

To clean it up, change the AVG(1) line opacity to 0% so it appears just as text! And then, of course, hide headers to get a clean viz.

Image and dashboard belo!