1 of 1 people found this helpful
You can always create measures from your dimensions such as datediff between two date fields (returns an integer value) or count of a dimension value.
2 of 2 people found this helpful
It all depends on the purpose of the Viz, the audience...etc. Sometimes it's not appropriate (or required) to make something "all singing and dancing" if the requirement is just to convey information in an effective way.
That said I think the below take on your Viz is a little quicker to ingest
I think the Filled/Unfilled shapes give a better (quick) view on the coverage, and I've also used a colour scale here. On the colours....as the Ship Mode goes "better to worse" it might be useful to visually indicate this...however in your real world example if the dimension (being represented by Ship Mode here) has no "good to bad" relationship it might not be appropriate to add colour (I try and only add extra colours, if it means something and/or aids understanding). Alternatively if that dimension has a "natural" colour (eg. if they were brands) using the Brand colour for each one would aid the reading.
Hope that helps
Countd() is your friend. Number of records often works, but I prefer Count Distinct as I explicitly know what the number represents, it also allows you to count things above the row level, for example, how many sub-categories are in each category, or which sub-category has the most products
if you create a calculated field (or use number of records): Number of orders = Countd([order id]), we can use the quick table calc: percent of total to get something like:
I work with several datasources that are dimension-heavy/only. There really is no limit to what you can do with just dimensions, you just have to create some measures first (I usually do this first and then publish a datasource to server with my calculated measures). I have attached a workbook with some other viz ideas/starters
Hope this helps!!
I've spent the last month working with dimension only datasets. It is interesting how much you can get out of them. And yes COUNTD() is your friend.
Thank you so much for your reply.
I know COUNTD() and DATEDIFF(), but I'd like some advice about how to visualize data only dimension, not how to create measure.
For example, Loliop chart seems to be more stylish than Bar chart.
Just like that, I'd like to make this matrix table more stylish.
If you want to keep the matrix table, but make it have more visual impact, play around with size, shape and color, for example, assuming the emphasis is on points that have more than 100 records ("A" in your calculation 2) then you could do something like this:
Since we don't care about about items where records<100 I have made them smaller and "grayed out", In this case it makes it very easy to see what the prominent ship modes are.
check out Shape Properties for how to add custom shapes (I grabbed a set of 1000 shapes a few weeks ago and sorted them in to folders so I would have some more options with out searching them out for a specific report)
Hope this helps!