LinkedIn
Twitter
As of version 3.28, you can now plot measure vs. measure scatterplots. This can be done by hiding the dimension(s) in the data table from the visualization.
For example, say you have this data table:
In order to chart Orders Count vs. Order Items Total Sale Price, you will want to hide Users Age from the visualization by selecting Hide from Visualization from that column’s gear dropdown:
After selecting scatter from the chart icon options, you should see a chart like this!
Hi, I have a hack that might help here. But please note the code is a bit ugly… Here is what I think you are trying to get to, measure by measure pivoted on a dimension.
Basically the problem is that you need to plot a measure on a dimension in a scatter graph if you are pivoting. So this dataset I am using here has a grouping, distance and a duration for each id. I want to plot distance by duration pivoted on grouping.
The solution requires me simply to plot distance (as a dimension) by duration (as a measure, sum is fine). However when doing this, we run into the problem that the distance could have duplications. To get around this, I have a new dimension called distance_for_scatter, where I simply add 0.0000001*row_number which will guarantee that the distances will ALWAYS be unique. (obviously add enough zeros!). Then you can change the value_format_name to something less horrific like decimal_2, and it plots nicely!
dimension: distance_for_scatter {
type: number
sql: ${distance} + (0.000001 * ${row_number}) ;;
value_format_name: decimal_2
html: {{rendered_value}} km ;;
}
If you don’t have a row_number dimension, you can change the sql_table_name to a derived table with something like:
SELECT *, ROW_NUMBER() OVER() as row_number FROM tablename
and then have row_number as a dimension, type number.
dimension: row_number {
type: number
sql: ${TABLE}.row_number ;;
}
Oh and change the minimum Y value to 0.00001
Mais voila 🙂
I’m also very curious about this.
Measure x measure scatter with dimensions is shown in Looker’s documentation here.
It’s the first image in the Grid Layout section:
@izzymiller could you share how this was built? Thanks!
FYI @AA_Sacco or anyone who has run into the same problem, I have been able to do this by converting the x-axis measure to a dimension in a new derived table while carrying the y-axis and name dimensions. Then in the explore of the derived table, pivoting on the name dimension that names each dot, so that the names come up as a legend. We can also size the dots in the visualisation options using the x axis dimension as long as you have the dimension declared as a type number. That should build the chart like the one in the doc.
PS: don’t forget to bind filter from the derived table to the source table
Joined to add to the mob.
I’m flabbergasted that the graphing abilities, and especially the scatterplot abilities are so primitive, having R scripts besides Looker to perform something so simple isn’t OK.
Can y’all hire someone from Tidyverse to clean up the graphing in Looker please 
