Today our instructions from Andy were as follows (these are copied verbatim from the original blog post) :

  1. The data come from the City of New Orleans Open Data Portal (here). They MUST use the API to get the data. They need to get all data for all years. This should be a simple task.
  2. They must use Power BI. 
  3. They are not permitted to use Tableau at all.
  4. If data prep is required, they can use either Alteryx or Tableau Prep.
  5. They are more than welcome to supplement it with additional data.
  6. Before 5pm, they must have their workflows complete, dashboards uploaded to the Power BI gallery, and blog posts written.

During the morning I was shadowing Ben Moss whilst he taught an Alteryx Essentials class so I was rather late to the party. This means I got Andy’s permission to skip the data prep and download a CSV file directly from the Open Data Portal. I downloaded Power BI Desktop and loaded the file directly in. This is done with the “Get Data” button. I have to say that Power BI and I were off to a bad start. I find the name “Get Data” extremely misleading as I already had the data I just wanted to bring it into Power BI. I presume this is because the button can be used to connect to various data sources including DB’s and local files, however, to get to CSV files I had to click on “More” at the bottom anyway so why not just make them separate buttons Microsoft and get the phrasing right?!

Anyway, I digress, I brought my data in and had a little play.

I soon found out why Power BI can prove to be a gateway into visual analytics for businesses and also why those who have found Tableau don’t tend to look back. Power BI strikes me as a prime case of one step forward, two steps back. For every feature I found that I felt was a marginal improvement over Tableau, I soon found several drawbacks and quickly understood why it wasn’t featured in Tableau in the first place.

A prime example of this is the canvas. You can insert charts anywhere. None of this containers business. “This is the life” you think to yourself, as you select each chart, go to format and then select distribute horizontally, until all your selected charts move to random points of your canvas with no discernible logic and you end up having to fit hem all manually. Then, heaven forbid that you change the shape of one as they WILL just overlap and you will have to go through it all again. It’s possible that I missed something as I only had a couple of hours to acquaint myself with the software, however, they don’t automatically move to take up available space. The feature in Tableau used to irk me in certain situations but, having seen the alternative, I’m actually pretty grateful for it.

Another feature of Power BI that I could understand people getting excited about was the automatic interactivity built in, however, I couldn’t find a way to customise what my actions on the dashboard did to the view and I soon found it to be more of an annoyance than an advantage as it meant that you had to be more careful with data aggregation etc.

You can’t customise tooltips.

You can’t select the order of aliases in the colour legend unless you separate them with if statements and then drag these calculated fields into different orders on the values pane.

Colour schemes are not maintained over the different types.

Depending on which Map chart type you select Power BI will aggregate your latitude and longitude your data differently. Infact, depending on the map chart type you are using it’s better just to use postal codes or state names rather than lat and long. as it will simply use an average lat and long. This makes lat and long pretty useless for this map so it kind of begs the question as to why these options are even included in the user interface.

The drag and drop features of Power BI pale in comparison to Tableau.

It’s almost 5 pm and I’ve run out of time to do an in depth analysis of the various features I’ve found. Overall, I think Tableau is easier to start with and you can go further with the tool. I really wish I could have wiped my brain of all data software knowledge prior to today as I can’t fully assess how fast I would have picked up BI if I wasn’t already familiar with Tableau as there are some similarities but I do think that Tableau is more intuitive. Some areas of Tableau (a great example is viz interactivity) can take longer to learn, however, this is mostly because there aren’t the options to customise these features in Power BI. On the other hand, the fact Power BI includes these automatically can lead to some more interesting looking charts straight off the bat but I think the decision to include these features in a chart should always be a conscious choice rather than a default as it reduces the chance of misleading charts.