(Updated 1/4/2012 – “Trends” (6 in a row either all rising or all falling) and “Shifts” (9 in a row either all above or all below the mean) – See Word doc in item 2 below, not reflected in the video.)

What’s next: (Done – 1/4/2012)
I would like to add “trends” (6 or more either all rising or all falling) and “shifts” (9 or more on the same side of the average line). This would require modifying the “Signals” and “MR_Signals” calculated fields to include these IF / THEN statements.

Any thoughts or comments? If you see an easier or better way to get the job done, please let me know!

Did anybody get back with you concerning adding “trends” (6 or more either all rising or all falling) and “shifts” (9 or more on the same side of the average line) to the Tableau Control Chart? I tried toying with it, but did not have much luck. Do you have any insights or programming suggestions as to what would be required to add trends and shifts to control chart?

Hi Ross – I haven’t figured out trends and shifts yet, but I’m pretty sure they will need to make use of the LOOKUP function to compare values to previous values. I’ll play around with it some more and ask some experts to see if anyone has any suggestions.

Hi Paul, thanks for the heads-up! Not sure what happened there, but I remapped the links to a dropbox location. It should be working now – apologies for the inconvenience.

Hi Paul – There is a “Size” slider in the Marks shelf that should help you size the circles. You have to make sure you have “Multiple Mark Types” chosen, and that you are editing the circle shape mark only (not “All” or the line). Hope this makes sense.

I’m brand new to Tableau but not to control charts, having built a number of them in Excel. Thanks for this lesson, I found it easy to follow and really helped me get to know Tableau a better. I happened to notice that the values the control limits in the chart I made in Tableau differed from the one I made in Excel using the same data. I believe the reason is that Tableau is trying to calculate the moving range for the first data point since I got a null value for this point. If you filter it from the moving range chart, it is also filtered from the control chart. If you include it in the data, it is assuming a moving range of zero, which affects the moving range. Either way would be inaccurate. Is there a way to exclude the first data point from calculation in the moving range but still have it plotted on the control chart?

Great Stuff Ben…one suggestion to potentially improve the Signals formula and it’s simple.

I would move the ‘outlier’ logic to the end of the IF statement. Here’s why: the UCL and LCL lines already indicate outliers. Let’s assume the 6th consecutive data point would indicate a trend in one direction or the other…in this case if that 6th consecutive data point would be colored as an “outlier” and you might easily miss the “trend”. since the UCL and LCL are already their to help show outliers, I would prefer this simple change.

just blogged on my site about setting the table calcs for UCL and LCL as well as signals to compute at Pane Across. After further review, I believe your Compute at Date accomplished the same. But the default for Compute at Window Across is no bueno.

Hi Milad. Yes, you should be able to create step changes in your control chart by simply adding the dimension (before/after) that delineates the different time periods, and the control limits will recalculate for the portions of the timeline in each period. Like in this example, where Peyton Manning’s stats are recalculated for the games with Indy vs the games with Denver: https://public.tableau.com/s/blog/2013/11/how-make-control-charts-tableau

I tried to reverse-engineer your Peyton Manning chart. I can get the averages to compute for each segment (either by team or by year), and the average moving ranges are computed for each segment, but the calculations for the limits all use the table average all the way across for the average moving range. How do I get the limits to use the average moving range for the segment, rather than the whole table?

Hi, It looks like the link to the Word Doc file is gone. I’m very interested in the trends functionality and was wondering if you still had it available? Thanks!

Ben,

Did anybody get back with you concerning adding “trends” (6 or more either all rising or all falling) and “shifts” (9 or more on the same side of the average line) to the Tableau Control Chart? I tried toying with it, but did not have much luck. Do you have any insights or programming suggestions as to what would be required to add trends and shifts to control chart?

As always, thank you.

Ross

Hi Ross – I haven’t figured out trends and shifts yet, but I’m pretty sure they will need to make use of the LOOKUP function to compare values to previous values. I’ll play around with it some more and ask some experts to see if anyone has any suggestions.

Thanks,

Ben

Just made the updates to add “trends” & “shifts” to the signals equation.

I’m having trouble downloading the data. Extracts to a series of xml files. I was expecting an xls file. What am I missing?

Thanks.

Hi Paul, thanks for the heads-up! Not sure what happened there, but I remapped the links to a dropbox location. It should be working now – apologies for the inconvenience.

All is well now. Thanks, Ben.

Question: How did you get your circle shapes for Chosen Stat to be so small? I can’t seem to size them to my liking.

Paul

Hi Paul – There is a “Size” slider in the Marks shelf that should help you size the circles. You have to make sure you have “Multiple Mark Types” chosen, and that you are editing the circle shape mark only (not “All” or the line). Hope this makes sense.

Hi Ben,

I’m brand new to Tableau but not to control charts, having built a number of them in Excel. Thanks for this lesson, I found it easy to follow and really helped me get to know Tableau a better. I happened to notice that the values the control limits in the chart I made in Tableau differed from the one I made in Excel using the same data. I believe the reason is that Tableau is trying to calculate the moving range for the first data point since I got a null value for this point. If you filter it from the moving range chart, it is also filtered from the control chart. If you include it in the data, it is assuming a moving range of zero, which affects the moving range. Either way would be inaccurate. Is there a way to exclude the first data point from calculation in the moving range but still have it plotted on the control chart?

Great Stuff Ben…one suggestion to potentially improve the Signals formula and it’s simple.

I would move the ‘outlier’ logic to the end of the IF statement. Here’s why: the UCL and LCL lines already indicate outliers. Let’s assume the 6th consecutive data point would indicate a trend in one direction or the other…in this case if that 6th consecutive data point would be colored as an “outlier” and you might easily miss the “trend”. since the UCL and LCL are already their to help show outliers, I would prefer this simple change.

Thanks again. See you in Vegas!

just blogged on my site about setting the table calcs for UCL and LCL as well as signals to compute at Pane Across. After further review, I believe your Compute at Date accomplished the same. But the default for Compute at Window Across is no bueno.

http://vizualinsight.blogspot.com/2015/09/tableau-control-charts-resetting.html

Thanks!

Do you by any chance know how to implement step changes for control chart in tableau. An example in excel can be seen here:

https://www.qimacros.com/control-chart/show-process-change-control-chart/

Thanks!

Hi Milad. Yes, you should be able to create step changes in your control chart by simply adding the dimension (before/after) that delineates the different time periods, and the control limits will recalculate for the portions of the timeline in each period. Like in this example, where Peyton Manning’s stats are recalculated for the games with Indy vs the games with Denver: https://public.tableau.com/s/blog/2013/11/how-make-control-charts-tableau

Thanks a bunch!

You’re welcome!

I tried to reverse-engineer your Peyton Manning chart. I can get the averages to compute for each segment (either by team or by year), and the average moving ranges are computed for each segment, but the calculations for the limits all use the table average all the way across for the average moving range. How do I get the limits to use the average moving range for the segment, rather than the whole table?

Hi, It looks like the link to the Word Doc file is gone. I’m very interested in the trends functionality and was wondering if you still had it available? Thanks!

Hi Chuck, thanks for the heads-up! You can go ahead and try the link now.