Your problem: You’ve got some map polygons in KML format and some data in SQL,Read More
BI tools – new to Tableau? You need to know about these unique features
This is the Tableau features list in the series of what I like in BI tools. Being such a popular visualization tool, I’m sure you as a new user will find many lists like this all over the internet. So I’ve made the list but also where and why I used, sometimes in a non-conventional way.
This is where you can call functions and create on-demand calculations based on your data. Useful for things like aggregates, chopping up string values or setting dynamic values based on a parameter selection. You can even send raw SQL to your data source from it to redefine a field as a different type.
But this is also where a lot of creative things happen. Because you can basically put anything in there and tweak a stock visual like a bar chart into a radial one. Admittedly, my experience tweaking other people’s examples to suit my own needs has been quite difficult. This is because of different data types, levels of granularity etc.
Some of my favorites I’ve seen online:
I found this very useful because there’s always a data type that prefers a particular chart type and this greatly expands the toolset. For example, when I worked on my TfL bus data, I wanted to show the most active bus routes. But the width of a bar chart can only show so many bars but a radial one can show much more.
Once you’ve got the hang of things and made a few sheets, you’ll probably put them into a dashboard and produce a story on them. But you’ll need to link them all together through Actions.
This allows the selection of one chart affect another and is effective for storytelling. For example, if your sales by store selecting a map geographically will filter the bar chart to only view data on that store. It’s quite versatile, allowing selections to be based on hovers and what to do when nothing is selected.
Recently they added visuals to tooltips so you could be more informed about what to hover and click on, especially if the data source isn’t particularly fast.
The issue is that as your project grows, you may want many of these and managing the web of actions can be quite difficult, especially if you decide to add a new sheet to your dashboard later. (It applies every action you’d made to that new sheet).
TabPy – Python in Tableau
TabPy is the ability to run the Python programming language connected to your Tableau project. This greatly expands what Tableau can do as it can send data to the Python environment for more complex processing not possible in the calculated field.
But what I used it for is looping. If you’ve used SQL, depending on which one you use, you know it’s difficult to loop. But with TabPy you can connect to your data source via something like ODBC (the speed doesn’t matter as we’re submitting commands not receiving any results) and push SQL queries in a Python loop.
This is great for underlying machine learning routines. For example, I had a clustering routine (expectation maximization) and this needs to loop until one of the parameters becomes extremely small. Obviously, you don’t really want to wait that long if you’re showing it to someone in Tableau but setting a hard limit for the number of loops is feasible. I don’t see much coverage on this functionality which is a shame considering the possibilities!
Things yet to try
I haven’t tried these yet but they are new features that you may want to be aware of.
Spatial joins – I’m glad BI tools in general are recognising the difficulty of joining location-based data.
Dual axis improvement – You can now plot polygons and independent location spot values together.
Things I’d like to see
Dynamic parameters – users have been asking for this for years and they’re currently working on it. It allows parameters, which take values from another field, to change accordingly. Very useful for dashboards with lots of new periodic additions.
Dashboard navigation – an upcoming feature. An often overlooked part of dashboard design is helping the user navigate it. You won’t be there to explain it to every user so navigation prompts will help them understand what they’re looking at and how to apply actions and filters.
Time series based visuals – There are line charts but they’re very basic. I’d like to see things like confidence interval plotting added so that I can see how well my forecasts are doing. Quite a specific request but I feel it’s one of the things I always have to go back to programming languages like R and Python to do.
Hopefully, this has been useful to new users who might be overwhelmed with the number of features, sometimes hidden away, in BI tools. Did I miss one of your favorite features? Or have I not discovered the proper way to do something in my improvements list? I’d love to hear about them and expand my BI knowledge.
Want to know how BI tools benefit work with Kognitio? Check out our solutions page for BI and analytic leaders. Also, check out this video on analyzing Transport for London bus data using Kognitio and Tableau.