As I promised you in the Complaints blog post that I’m going to unfold the steps of creating the dashboard itself, so here it is!
The above is a first prototype for the Complaints dashboard that served perfectly well when I had to create the dashboard itself.
I highly recommend you to grab a pencil and draw something first, since it is much more easier to create a higher level version of the notion you have in mind, than putting it all together at once in Tableau. You can have a rubber anytime and fix the errors with one move.
The paper prototype also lets you to share your thoughts easily with your collegaues. It facilitates collaboration and groupwork as well.
Connect to… everything
With Tableau you get an exceptional BI tool that lets you connect to whatever you need to connect to. When you start up Tableau you can see a wide range of databases on the left side of the connection panel (the range depends on which version of Tableau you are using).
Select MySQL from the list to connect to your local database.
You will have to give credentials for the connection – Tableau simplifies this process as seamless as possible.
If you select your table from the list (currently there’s only one, but later… you will find more) Tableau asks how much data shall it use. It’s worth thinking about the decision since when you have small databases for example, than you might load all the data into Tableau, but when you have a huge database system with many tables and you run simple queries against it you might connect live to your data source.
Building the worksheets and the dashboard
Once you are connected to the MySQL data source, the columns of your table should appear as dimensions and measures in Tableau. The tool has many usability advantages, like it tries to understand your data type and rates it to either dimension or measure category. Later on you can change these by simply dragging your data pill into the other category. For example you need to change the dates to date format by right clicking on them and select „Change data type” -> „Date”.
Since you have the overall look already at hand, you can start creating the different worksheets based on that inital drawing and set up the final dashboard.
Creating the worksheets
Creating a diagram is the easiest in Tableau (one of the reasons why it is a popular BI tool). You can either drag and drop measures and dimensions to the rows or column panel, or simply double-click on a dimension and measure, and Tableau figures out the best diagram type for you.
Clicking on the „Show me” tab in the top right corner reveals the different diagram types Tableau currently capable of. The ones that are grayed out are not available for the current measure/dimension set, but if you hover over a specific diagram you can figure out what it needs to show.
The line graph
So, accordingly, grab the „Date received” to the column panel and the „Number of records” to the row panel to create the line graph that shows how many complaints were at a given time.
So now you get a fine line graph.
One thing to mention regarding line graphs is that the slope of the graphs should be around 70 degrees for maximum legibility of the graph (Jacques Bertin: Semiology of graphics, ESRI, 2010.).
The stacked bar graph
Using the same drag & drop technique you can create the stacked bar graph which shows the distribution of complaints by company and product/service type.
Drag the „Number of records” pill to the column panel and the „Company” pill to the row panel. Then drag the „Product” pill to either to the detail mark, or you can color the stacked chart if you drag the „Product” pill to the „Color” mark and so there it is.
You will also need to order the companies by complaint number, which is easy to do by clicking on the order by icon at the top of the graph. By now you have the final graph, hopefully similar to the below one.
Maps with pie charts
There are four different maps combined with pie charts that show distributions of complaints by different measures. I’m not going down too deep into the details of creating a dual axis map with pie chart, since you can read a really good article here about it and it explains everything for sure.
The main issue with any map + pie charts type diagram is that size of the pie charts. If the pie charts are too small, you won’t be able to see the different slices of the pies, if they are too big than you won’t see the underlying choropleth map (colored by a measure). The already cited french cartographer and diagram theoretist states that the pies should consume only the 10% of the whole combined chart. You can resize these by clicking on the „Size” mark and drag the slider to adjust the size.
The final results should look similar to this:
Note: when you are in worksheet view, the graphs look different when you are in dashboard view, since you are going to resize these to fit your original screen size (that you already determined hopefully).
When you have repeating patterns in your dashboard it’s quite time saving if you create the first version of the diagram pattern and then simply duplicate it as many times as many unique diagram is needed and, drag the new measures/dimensions on the top of the old ones replacing them like this.
Creating the dashboard
When you have your worksheets ready you can create the dashboard itself. Add a dashboard tab and you can drag the sheets to it. But wait, these are rearranging themselves, is it ok? Sometimes. If you don’t want the sheets to be automatically rearranged click on the „Floating” button on the middle left, so you can align your text and graphs pixel perfectly.
You already made a draft how your dashboard should look like on paper, so you can use it as a starting point when building the actual dashboard. You don’t need to stick too heavily to the prototype, it’s only there to help you, not to restrain you.
You will also need to add a few details, the title, subtitle, footer in case you need to reference your datasource (most cases) and some other small polishing to get the dashboard done, but that will be the topic of another post.
You can have a look at the final dashboard at Tableau Public:
Looking for part I. of the Complaints article?
Did you like this article? Follow us on Twitter!
- Developing a simple dashboard in Tableau (part II.) - October 13, 2014
- Creating a simple dashboard in Tableau with Talend and MySQL - September 15, 2014