Period over Period (PoP) Chart Labels


Have a simple question that may have been discussed before so my apologies in advance for asking again :)…(actually had a ticket on this from February 2019 but this question came up. again for us). I am using PoP analysis (OFFSET filter functionality) that is built in Dundas (version 7.0.2) but our users want to see the LABELS on the graph (see attached - first screenshot). Currently, all I can show for PoP for last 12 months on the graph is Current Period and Previous Period for either Series or Legend LABELS. So for example if I am doing June 2020 - May 2021 for Current Year that date range will show on my X-axis and SERIES/LEGEND LABELS will show Current Period and Previous Period (technically that would be then June 2019 - May 202) if we reference example provided). So our users want couple of things (and they are sticking hard by that request :)):

  1. Labels showing what we are comparing (YoY, MoM, QoQ - based what we select in OFFSET filter) AND/OR
  2. Showing (referencing this example) current month and previous month SIDE-BY-SIDE (so June 2020 and June 2019 bars would be next to each other) - maybe somehow in screenshot #2 I attached which is NOT using built in OFFSET filter. Third screenshot shows that this is possible in Excel and I believe in SSRS as well.

I was thinking maybe a potential solution IF we continue using OFFSET filter v. say 2 Calendar Range selections could just be adding a ‘dynamic label’ on top showing what is Current Period and what is the Period that Current Period is being compared to (i.e. Previous Year in this example), however, I think that I’d need it to be DYNAMIC (i.e. if say LAST 4 Quarters is selected in OFFSET menu or LAST 90 Days, etc.)? So thinking in my example 2020 and 2019 would be sufficient as LEGEND LABELS? If this suggestion could work, would appreciate any info on scripting this out somehow.

Hope I am not being confusing here.

Thank you!

Hi Semir,

I think that when you are comparing periods like year-over-year in Excel you probably have this data in separate columns/series already, so that each series can be identified in the legend normally. You can do this as well in Dundas BI by putting the time dimension on Columns in the Data Analysis Panel, and if there is a Hierarchy Value filter connected, users could pick and choose individual members such as Dec 2019 & Dec 2020 using checkboxes and those dates would then be shown in the legend like in your 3rd screenshot. A possible difference here is that the time dimension is no longer on Rows or along the X axis. Another option is to put the time dimension on Slicers and use a slicer comparison to provide two filters that can each be set independently to any two time periods to compare.

You could still put something else like the time dimension attribute Month of Year on Rows & the X axis while doing this to look at the monthly data while also comparing year-over-year, for example (or a non-time dimension like State in your 3rd screenshot) but otherwise Dundas BI’s Period Over Period feature has the most flexibility for combining grouping by time periods with any kind of offset to be displayed alongside them. When using the actual Period Over Period feature, the offset (Time Lag) filter itself is providing a description if it’s provided to users, and I’m not sure exactly how you would want to go beyond this. In your examples, “previous year” does not seem accurate and you couldn’t just label 2019 vs. 2020 in the legend because the monthly data begins and ends mid-year for both series, so you’re really looking at “-12 months” or “12 months prior”, more closely resembling the offset filter. If you did decide you wanted to calculate some exact dates to display, the hierarchy members in the cellset shown in visualizations and the member values set on parameters each have a memberTime property with the corresponding dates (the member values on parameters are offset to UTC time).

1 Like