Forum Discussion

Anonymous's avatar
Anonymous
Not applicable
4 years ago
Solved

Recommended visuals for show Distributions and Standard Deviations?

EDIT 11/10/21: Please note that I say some factually incorrect information in the below post. I am new to Power BI and the solution provider was kind enough to educate me on my misunderstandings. I honestly recommend just scrolling down straight to the solution and looking at their .pbix file. It helped me and might help you if you, like me, were struggling to find a scatter plot or other visual with a dynamic standard deviation line.

 

What are recommended visuals for show Distributions, means, medians and Standard Deviations? I notice that one of my stakeholders very much likes to look at standard deviations, specifically to see what items are within two standard deviations, what items are outside of two standard deviations, etc.

 

I am aware of the below options for showing distributions:

- Histograms: The issue here is that while I'm aware histograms can be manually divided into bins, I'm not yet aware of a histogram that dynamically divides into standard deviations. 

 

- Scatterplots: I am aware that you can use the default scatterplot visual analytics pane to add lines to the default scatterplot visual for means, medians, percentiles and X/Y-axis constant lines. However, none of these lines accept measures as input--you have to manually pick a number for the constant for example, so that prevents me from inputting a measure that could create a dynamic standard deviation line.

The percentile could also stand-in as sort of a replacement standard deviation assuming that your distribution's overall shape doesn't change. However, it's not a proper standard deviation.

 

- Line and clustered column chart: I like what I have seen people do with these (ie adding a line for a standard deviation measure, a line for a mean measure, a line for mean + standard deviation). I think column charts work really well when you are looking at sums (ie a sales team's total sales by individual for example, so you can quickly see how much each individual sold as well as their mean and standard deviation). 

However, in my case, I am wanting to visually show data points that represent cycle times, and it doesn't seem quite right to sum up one individual's cycle times.

 

- Box and whisker chart: I think this is also a good option that provides an IQR and allows you to see where different data points land in a distribution, but it still doesn't provide the standard deviation. My boss also does not think enough people understand what a box-and-whisker chart is for it to be useful.

 

Am I overthinking my problem? Does anyone have advice? I made some dummy data below to give an idea of the kind of data I am working with:

 

SportGoalSession DateSession IDTime (minutes)Athlete
Climbing305/1/20211837Natalie
Running105/8/20211920Ryan
Climbing305/8/20213054James
Running105/9/2021412Natalie
Climbing305/10/20214530Natalie
Running105/19/2021124James
Swimming455/23/20214444James
Swimming457/10/20211745James
Swimming457/4/20211146Ryan
Swimming457/19/2021546Natalie
Climbing306/10/20212147Micah
Swimming4510/13/20212046Natalie
Climbing306/20/20214234Natalie
Running107/1/20213114Micah
Swimming4510/5/20211448Ryan
Swimming455/21/20212349Beth
Climbing307/13/20213639Son
Climbing307/15/20212446Ryan
Climbing307/16/2021945James
Swimming459/2/20212951Beth
Climbing307/23/20212735Ryan
Climbing307/29/2021343Micah
Swimming4510/22/20214151Natalie
Running108/4/20211611Natalie
Climbing308/16/20213940Son
Running108/20/20211320James
Climbing308/22/20211537Son
Running109/1/20212821Natalie
Swimming456/12/20213255Beth
Running109/3/20214014Micah
Running109/6/20211011Ryan
Running109/6/20213425Son
Climbing309/14/20211239Beth
Climbing309/15/2021639Ryan
Swimming459/20/20212655Natalie
Swimming456/6/20213560James
Running1010/1/20214321Natalie
Running1010/4/2021724Micah
Swimming456/2/2021861Son
Running1010/6/2021229Micah
Running1010/8/20213720James
Swimming459/21/2021263James
Running1010/19/20212520Ryan
Climbing3010/22/20213350Beth
Swimming458/3/20213864Son

 

I can also upload a sample .pbix showing dummy data where I've played with standard deviations--I see others have done this on here but I'm not quite sure how to upload it myself.

  • The constant lines can take measures in their definition.

     

    Here's an example based on your data. Each of these lines is defined by a measure.

     

    See the attached file for more detail (and try clicking the pieces of pie to see it dynamically change).

2 Replies

  • The constant lines can take measures in their definition.

     

    Here's an example based on your data. Each of these lines is defined by a measure.

     

    See the attached file for more detail (and try clicking the pieces of pie to see it dynamically change).

    • Anonymous's avatar
      Anonymous
      Not applicable

      This was so helpful! Thank you so much for teaching me this. I had no idea previously.