QlikView vs. Tableau

Some while ago I was asked by the CIO of my company to give a report on the Microsoft BI tool set. While gathering information I constantly found QlikView and Tableau being named as the new top dogs. Having used QlikView for a couple of years now I got interested to know more about the rival.

Just by reading blog posts and articles it already became clear Tableau fits into the category of data analysis tools. QlikView emphasizes on data discovery, so would it be fair to compare them against the same criteriors?

Planning to buy me a new car I needed to analyse the balance history of my bank account. This gave me a perfect scenario to try to implement using both tools utilizing the data I already had ready in a Excel workbook.

QlikView

In QlikView the building of a solution always starts by configuring the load script defining how the data will get imported. QlikView is never directly connected to a data source, instead the information is always first imported, processed and compressed.

I had to add two data transformations: convert the date information to a correct date format and to change the numbers with a trailing negative or positive sign to the more traditional format. To help with displaying the values on a monthly and weekly timeline I also had the script generate some additional information.

QlikView load script

When I was ready with the importing, setting up the chart and table objects was easy. As I only had data from the beginning of this year I configured the accumulation “manually” by defining it in the chart expression enabling me to include the initial balance.

QlikView solution

The final result looked nice and as QlikView by default already supports filtering by clicking on any visible information I did not see a need for any additional filter boxes.

Tableau

My plan was to use Tableau Desktop but as my trial license had expired I had to turn to Tableau Public that is offered free for publishing dasboards online.

Configuring my Excel files as a data source was easy using the inbuilt wizard. Unike QlikView Tableau links in general directly with the data source and the transformations are processed on the fly.

As the data required some tansformation as desrcibed earlier, I created two additional calculated dimension fields for this. The script language was quite close to what I used in QlikView so I could copy-paste most of what I had earlier written.

Creating the list was easy and done by drag and drop just like when working with pivots in Excel. Configuring the chart was more of a challange as I needed to create some additional calculated measure fields to get the same result I had in my QlikView solution. In contrast to QlikView, I did not have to create any week and month fields has Tableau could do the groupping directly using the available date field by selecting the desired option from the menu.

Tableau workbook

In Tableau the visual elements are defined in worksheets and these are then composed on a dasboard. Setting up the dasboard was easy and done by placing the previsouly created worksheets onto a grid. As there is no standard drill down and filtering available the presentation ended up being quite static .

Tableau dashboard

Summary

Using QlikView, writing the import script and configuring the objects on the canvas requires some degree of knowledge but it also allows the user to setup complex data handling and having full freedom to define the visual appearance of the solution. QlikView enables not only setting up dashboards but also has potential as a platform for creating complete BI tools.

Despite the fact that I did not spend any time on reading manuals or watching video tutorials before getting started with Tableauwas easy. Tableau also offered much of the functionality through menu options that required scripting in QlikView. Unfortunately when working with dashboards what you see is what you get, leaving less room for discovery.

I would imagine Tableau is closer to Microsoft PowerPivot in its user experience and approach than to QlikView. As there tools are of different nature, only by comparing them against specific requiremets will yeld in a truthful conclution.