Measure vs. Measure Scatterplots (3.28+)

(Mark Goodwin) #21

+1 for having the dimension name in the tooltip. Otherwise, I can see the outliers but don’t know what they are! Plus I would want to be able to drill from that dimension from the scatter plot in order to get more information about it!

My scenario is - I am looking at communications sent by the marketing team, and wondering on which device they are getting read - would like to scatter the number read on the mobile v. desktop then hover over the data point and see which email it is. Obviously, if more people are reading it on mobile I want to know that. And then click on the data point to see more information about this communication.


(Chris Seymour) #22

Hi @Mark_Goodwin,

I’ll let the product team know that you’d like to have the dimension name in the tooltip. As for drilling down by clicking a data point, that option is available! You’ll want to use the drill_fields parameter in the dimension that uniquely identifies your emails:



+1 for this! Hugely useful to know what dimensions the plotted points correspond to in a measure vs measure scatterplot.

(DCL) #24

Thanks for the upvote @Avocet! I’ll pass that on to the product team :slight_smile:

(Sean Higgins) #25

Some custom html and Liquid might help with some of the tooltip requests:

Here’s a simple example but could be expanded and made more fancy:

  measure: order_count {
    description: "A count of unique orders"
    label: "{% if users.age._is_selected %} Details {% else %} Order Count {% endif %}"
    type: count_distinct
    sql: ${order_id} ;;
    html: {% if users.age._is_selected %}
    <font color=red>Age:{{ users.age._rendered_value }}</font>, <font color=yellow>Order Count:{{ order_count._rendered_value }}</font>
    {% else %}
    {{ order_count._rendered_value }}
    {% endif %};;

When Age is selected:

VS when it is not:

(DCL) #26

Great add here @shiggins! This is another good post as well to show other fields in the tooltip using our rendered_value type liquid reference.

Here it is used to concatenate a percent of total value and the string “of total” to show a nice additional data point in the hover without affecting the ability to plot the original Total Gross Margin measure values

(Dimitri Masin) #27

I also think it would be super powerful to be able to segment by color e.g. by adding a pivot dimension. Right now the scatter plot is very limited. Are there any changes on the roadmap?

(Ryan Dunlavy) #28

Hey @Dimitri_Masin,

Tweaks to scatterplots including shading by color are not on our near-term roadmap right now, but I will let the product team know you would like to see this feature as well!


(Steve Caldwell) #29

@ryan.dunlavy - Huge +1,000 to being able to customize size, shape and color of scatter plots using dimensions or measures. This is a key feature in many other visualizations and is essential to getting the most out of a scatter plot - with color and shape to spot meaningful trends between different cohorts and with size to avoid letting low sample observations skew your interpretation of a data trend.

Please, please introduce this functionality soon!

Meanwhile, anyone have the best tips on preferred custom viz tools and options?

(Mani Popuri) #30

@ryan.dunlavy - +1, We really want to see the improvements (segment by color using dimension, and custom size to spots) on Scatterplots.

(Josiah Vincent) #31

I agree with @scaldwell and @mpopuri ; these are key components of a functional scatterplot. They are defined by the two+ measures and the ways to segment the view with color, size and shape. Until scatterplot has this functionality, there really isn’t much we can do with this plot option.

(Ryan Dunlavy) #32

Hey @Dimitri_Masin, @scaldwell, @mpopuri and @Falreign,

In Looker 5.18, as part of the Enhanced Visualizations labs feature, the coloring of points will be possible in scatterplots. This can be done multiple measures or pivots. See below for an example, and be sure to check out the full 5.18 release notes!

As for adjusting the size on scatterplots, I have pass all of your feedback on to the product team. Cheers!

(Josiah Vincent) #33

Oh, I see! Yes colors is the big one so I am glad to see it is here.

Thank you for this!

(Steve Caldwell) #34

@ryan.dunlavy - This isn’t a true scatterplot. A scatterplot graphs a mark for each individual record (no aggregation) on two particular values - these are typically both measures that won’t be aggregated. If looker needs to treat both as a dimension value to do this effectively that’s fine, but the Y axis should be another continuous value plotted against age, with the same number of dots as there are User Count for each matching value of age and second dimension.

What looker calls a scatterplot is just a line graph without a line. It’s not very useful.

Also, shape and size are equally important in this kind of data mining exploration. I would really like to see this functionality get better. Please add full scatterplot functionality to Looker.


+1 for a fully functional scatterplot. I agree with @scaldwell that this isn’t a true scatterplot.

For example, I want to measure # of incidents worked vs an escalation rate by teams. So my x- axis is # of incident, y-axis esclation rate, and the dots represent each team. The teams should be individually colored - so each dots is correlated is a value from a dimension within LookML. I don’t believe this is an option in Looker currently.

(Department of Customer Love) #36

Hey @dan.geneczko,

Good news! When using the Enhanced Visualizations labs feature it’s now possible to create scatterplots with various coloring for each series, or dimension value.

(Milan Veverka) #37

Hi @rachel.johnson , I was able to control the size of the “bubbles” (great), but couldn’t figure out how to use the dimension as a series (in a simple two-measures, one dimension case)… Any tips?

(Department of Customer Love) #38

Hey @Milan_Veverka,

Pivoting on the dimension will allow for a new series for each dimension value!