1 2 Previous Next 19 Replies Latest reply on May 3, 2017 6:31 AM by isa.serah

    Scatter Plot with dimensions

    isa.serah

      I have a requirement .

       

      We have a field called product category. I have to assign different level of criticality to each category as HIGHLY critical, Moderate or Low critical.And our idea is to create a scatter plot where in the X axis,we have to show the criticality level .

      X axis(criticality level)

      ----------------------------------------->

      low     moderate   high

      1           2            3

      Y axis

      There are different stages in the products.

      1.End of Life stage

      2.End of support stage

      3.End of extended support stage.

       

      Y axis

      ^

      |End of Life(1)

      |

      |End of support(2)

      |

      |End of Extended support(3)

       

      All these 3 are date fields showing when the product will be in that stage. If we could assign a number to each stage(for ex:End of Life -1,end of support -2,end of extended support-3),Then in the Y axis we can show how many products in each of these stages.

      So basically, I need a scatter plot showing who many products are there in all of these three stages(Y axis) and what is the criticality level(X axis) .Is it highly critical,moderate or low ?How will I do it since all these fields what I told about is dimension.?

       

      X axis(criticality level)

      ----------------------------------------->

      low     moderate   high

      1           2            3

       

      Y axis

      ^

      |End of Life(1)

      |

      |End of support(2)

      |

      |End of Extended support(3)

       

      Can someone help me ?Simon Runc

        • 1. Re: Scatter Plot with dimensions
          Simon Runc

          hi Isa,

           

          That sounds do-able...on the product life stage, is the there a tag/dimension to indicate which date is for which life stage? or do we always know that the first date is the End of Life, the 2nd the End of Support...etc.? and is there always 3 life stages for every product?

           

          If you can dummy up some data too, then I can look at an exact solution for you

          • 2. Re: Scatter Plot with dimensions
            Vincent Baumel

            The first visualization that comes to my mind isn't a scatterplot at all, actually. If you're visualizing the number of products at each intersection of a 3x3 grid, you should consider a highlight table rather than a scatterplot. Scatterplots can be very effective with certain kinds of data, namely continuous measures that may fall anywhere on a given scale (think things like Sales, or Quantity). When the entire range of possibilities is limited to 3, a scatterplot probably isn't your best bet. A highlight table lets you visualize counts at a given intersection of dimensions. If that isn't clear, consider a data set that looks like this:

             

             

            If we tried to make a scatterplot of this, we'd end up with a 3x3 grid that has multiple marks stacked on top of one another wherever they have the same configuration of criticality and stage. This makes it hard to see the number of items. Instead, if we make a highlight table out of it this is what we end up with:

             

            This is super easy for anyone to understand, and it immediately draws your attention to the distribution of values. Tableau makes creating a highlight table really easy, too - it's right in the Show Me pane. Simply ctrl+click your dimensions and measures, then pick the highlight table.

            4 of 4 people found this helpful
            • 3. Re: Scatter Plot with dimensions
              isa.serah

              No these date will be applied only to those products which are going to be expired.There will not be  life stage for  every products.A manager looking to the scatter plot has to understand ,these are the products in his account which are in the End of life period.(or end of support or end of extended support).

              There is no separate dimension to indicate  which date for which life stage..We have 3 date fields ..one for EOL(only dates),one for end of support  and other one for end of extended support.When we drag Product into rows and all the three dates together with it to rows,then we see many null values for most of the products..Only some products will have values for the date fields.

              • 4. Re: Scatter Plot with dimensions
                isa.serah

                Makes a lot of sence..This is simple and very communicative .Client asked to get a scatter plot for this. That's the reason behind this question..

                • 5. Re: Scatter Plot with dimensions
                  isa.serah

                  And we don't have a field indicating the stage of product.Just  date fields for each stage.no separate dimension telling stage for each product.

                  • 6. Re: Scatter Plot with dimensions
                    Vincent Baumel

                    Hmm. Well there are certain things you can do if all you've got to work with are date fields. If it's something like "All products aged between 30 and 60 days are considered end of life" then you can do some pretty simple calculations to create a Stage field. It depends on your data though.

                     

                    Allow me to wax philosophical for just a moment. I'm reminded of the times I had to defend my stance on 3-d pie charts that someone INSISTED I incorporate into the dashboard.

                    Your client may have asked for a scatterplot, but we need to diplomatically stand our ground as data visualization practitioners. If we know the message our clients are trying to communicate (e.g. "How many products do we have that are near or past their support?") then it is our responsibility to make the most solid case we can for the visualization that tells that story most effectively. We want the same thing the client wants, once you strip away the fine details of what graph types, color scheme, etc. they seem to have their minds stuck on. Your client may want a scatterplot, but not really have any idea WHY they want it. Or it could be that they are communicating it to you because it was communicated to them first. Either way we have an opportunity to set the bar nice and high for informative, engaging, and inspiring data visualizations. My integrity as a data visualization practitioner depends on it.

                    2 of 2 people found this helpful
                    • 7. Re: Scatter Plot with dimensions
                      isa.serah

                      Totally agree with you.....Thought of posting this question in the forum before deciding on the other views....If we cant do it as scatter plot, will go with other views .

                      • 8. Re: Scatter Plot with dimensions
                        Simon Runc

                        hi Vincent,

                         

                        I couldn't agree more...I stay away from letting the client have a say in which Visuals they want...after-all part of what they are paying for is our "Data Visualization" expertise. I just want to know what their problem is, and what levers they have to affect that problem...then it's up to me to come up with the best Viz. There is something about the visual fields (such as Data Viz, Graphic Design...etc.), where the language is accessible (Pie, Blue, Red....) that means everyone can make a (seemingly) plausible suggestion.

                        ...if they do try down this path I ask them if they could also help me diagnose the cause of the "hetroskedacity" of the statistical model I'm working on...[room falls silent]...I then inform them that I've spent equally as long studying statistics as I have data visualization! (very politely, of course)!...I also inform them that I wouldn't dream of suggesting how they should structure their HR department, as "I wouldn't have a clue what I was talking about"!!!

                         

                        If needs be I'll work up both options and demonstrate why one is better.

                         

                        Isa, On the problem at hand...I think the issue is getting a mark for each date (at -1 and/or -2 and/or -3) when they all reside in the same row (regardless of it being a scatter or a heat map). Are you able to Pivot the data?

                        1 of 1 people found this helpful
                        • 9. Re: Scatter Plot with dimensions
                          Vincent Baumel

                          This scatterplot would actually be a decent example of predictable heteroscedasticity :-D As criticality increases, the range of coverage is likely to grow as well!

                          1 of 1 people found this helpful
                          • 11. Re: Scatter Plot with dimensions
                            isa.serah

                            Simon,

                             

                            Yes ..that's the exact problem I am facing...Since all the three are different date fields, I am unable to assign a number to each date field. Stuck there completely..if we could assign a value to each stage then we can put as y axis.I can create a cross tab(pivot) showing products and its corresponding dates.Most of the products have null values in the date fields(as they are new products).

                            • 12. Re: Scatter Plot with dimensions
                              Simon Runc

                              So I've dummied up some data...so we can start investigating the problem.

                               

                              It looks like this

                               

                               

                              In Tableau I then Pivot this data

                               

                               

                              and after re-naming the fields it looks like this

                               

                               

                              From here it's fairly easy to assign numbers to each of the 2 measures....

                               

                              so I first create this

                              [Criticality INT]

                              CASE [Criticality]

                                  WHEN 'Low' THEN 1

                                  WHEN 'Medium' THEN 2

                                  WHEN 'High' THEN 3

                              END

                              and then

                               

                              [Status INT]

                              CASE [Status]

                                  WHEN 'End of Life' THEN -1

                                  WHEN 'End of support' THEN -2

                                  WHEN 'End of Extended support' THEN -3

                              END

                               

                              I could have done this in the [Status INT] calculation, but I just added this filter to the Viz, to remove any NULL dates

                               

                               

                              As you'll see, as per Vincent's advise Product 5 is hidden below another product (you can click on the legend on Product 5 to see it)...with a heat map you could see the number of products at each point on the matrix, and then (maybe) a click to a detail Viz detailing out the Products in each Status/Critical combination.

                               

                              Hope that helps.

                              1 of 1 people found this helpful
                              • 13. Re: Scatter Plot with dimensions
                                isa.serah

                                Simon,

                                 

                                I am unable to download the workbook since its in tableau 10 version.I have tableau 9.3 in my system.

                                • 14. Re: Scatter Plot with dimensions
                                  Simon Runc

                                  Here you go...Tableau 9.3 just survived my last Tableau version cull

                                  2 of 2 people found this helpful
                                  1 2 Previous Next