Overview

Tableau is a great visualisation software, arguably the favoured visualisation tool among data analysts. It's reasonably user friendly and you can create dynamic and interactable graphs that can provide a much richer experience to the user than a standard Excel plot can or PowerPoint can. As a reasonably new user to Tableau I wanted to document my experience with the software and also demonstrate how to create eye catching dashboards.

The dataset used in this article is provided from Kaggle link here: https://www.kaggle.com/datasets/mayureshkoli/best-universities-in-the-united-kingdom and is taken from the following sources:

1) https://www.ukuni.net/universities

2) https://www.4icu.org/gb/universities/

3) https://cwur.org/about.php

The main factor utilised in this dashboard is 'University Ranking', this is an aggregated value and is judged on several factors including teach to student ratio, amount of research conducted and so forth. More information about University rankings can be found here: https://www.timeshighereducation.com/student/advice/world-university-rankings-explained

Connecting to Data

Tableau has useful ways of importing data for visualisation. In an industrial setting with many different databases it is useful to connect directly to the database you want to analyse. However, in this scenario we've downloaded an xls file from Kaggle.

Importing data into Tableau

Tableau automatically detects categorical and continuous data types from our imported dataset and displays them in our Tables view.

Tableau automatically sorts our imported data

Ranking Universities

The first thing I want to do is to rank order the universities. Most things in Tableau are done by dragging our data categories into either the rows or columns table in the main worksheet.

University rankings in accending order (the top 10 are grouped)

Dragging 'UK rank' Into columns and 'University name' into rows creates the above graphic. We go into the details pane next to 'University name' to sort the data in ascending order. We select the field name as the data we wish to sort by, in this instance 'UK rank'.

How to sort data in Tableau

Geospatial visualisations

One of my favourite features about Tableau is its ability to easily create geospatial visualisations. When plotting the longitude and latitude, Tableau automatically imprints a map visualisation in the backdrop of what is essentially a scatterplot.

Plotting longtitude against latitude in Tableau creates a geospatial plot with a map background

We can add more to this plot now too. When we drag our University rank parameter to 'colour' in our Marks panel we can get a better idea of the countries top performing universities based on region (with green being the top performers and red being the worst).

Adding a third dimension to our plot using colour

I also decided to add text labels so you can zoom into the map to see your desired University and its ranking.

Adding labels to plots helps make them more interactive to users

Analysing Rankings by Region

The map provided some insight into rankings by region. It appears that there is not an obvious trend between rankings and region (for example if we were to make a scatter plot of latitude vs rankings we would not see a trend where the rankings increased the further south we moved. Interestingly however there does appear to be high ranked universities in the same city as low ranked universities. In fact London has universities in both the top 10 and bottom 10 of the university league tables. This is perhaps unsurprising given the size of the city of London, however similar trends appear elsewhere (such as with the University of Lincoln - ranked 51/131 and Bishop Grosseteste University ranked 102/131, and also The University of Leeds - ranked 13/131 and Leeds Beckett - ranked 123/131.

Let's see if there are more regional differences, we can easily plot the median ranking by the region to create a bar chart. Tableau automatically aggregates our data into the sum initially but we can change this to whatever aggregate function we want.

Median rankings by area

Changing aggregate functions in Tableau

Analysing Correlations

It looks like the midlands and Scotland have lower median rankings than other Universities from this chart. Though I'm curious to know what other factors influence a universities ranking. Let's take a look at the average annual student fees for undergraduates and create a scatter plot against the university ranking.

Scatter plot in Tableau

We can add a trend line to this scatterplot which automatically gives us some useful statistics, (note that I inverted the Y axis to create a more visually appealing plot, this can be done in axis options). Well, it looks like there is some correlation between the average university fee's and the rank it receives. This stands to reason when we consider that one of the factors that in scoring for ranking is teacher to student ratio and a university with more income from enrolments can hire more teachers. However, an R-squared value of 0.45 is pretty low. Less than half of the variation in the dataset is caused by the relationship between these two variables. We can see from the data however that there are some obvious outliers (Oxford Cambridge and Imperial College London are all skewing this value quite a lot, there is also Harper Adams university that stands out on the left of our plot (ranked 17) that has a very low enrolment fee.

We can simply click on these datapoints to exclude them from our analysis.

Removing outliers

Our new scatterplot with outliers removed

Our new plot looks quite different. There are still outliers but we don't want to adjust the data too much to fit any kind of bias, this is unethical. We can see from removing these 4 datapoints however that our R-squared value has increased by almost 5% more. So it does look like universities with higher enrolment fees have better rankings. But there are other factors at play than simply fees. There is a medium correlation between these variables.

Custom Formulas

Now I want to pull all of these graphs together into a dashboard to display the information coherently. We previously created worksheets for individual datasets, but we can create a new dashboard sheet and then drag our tables into it. Before I do this however, I want to create some KPIs/BANs to display. Let's list the highest ranked university, the most expensive and the oldest.

To do this, we need to create a custom formula, let's look at the most expensive university by admission fees.

We do this by going to analysis > create calculated field and then entering the following formula:

Custom formulas in Tableau are found by going to analysis and create custom field

We're creating a filter of the maximum value in the column 'UG average fees (in pounds) and setting it equal to itself. This creates a filter we can now use.

We can now drag our new field 'Highest Fees Filter' into the filters tab, this returns a single value, the highest value in this column. But we also want to know which university this is. So we drag our university names column to label to give this information.

Our finished KPI/BAN

Note the 'SUM(UG average fees)' is added when creating an annotation which we can see to the right of our results. This is done by: right click, annotate, mark.

The same steps as above are repeated for the oldest university and highest ranked university. We can now bring all of these elements together into a dashboard.

Creating a dashboard

The completed dashboard

The completed dashboard compiles all our our previously generated plots into a single sheet. Each is still a dynamic graphic and allows for easy exploration of the data. This is a great way to share visualisations within an organisation.

Conclusion

This data provided some insight into UK universities through data visualisations made in Tableau. We analysed the top 10 universities and also explored some of the factors that influence university rankings; enrolment fee's appear to play a part, and to a lesser extent so does location. If you're interested in exploring the dashboard I created please click the link below.

Thanks for reading, please feel free to connect with me on LinkedIn, or checkout my GitHub or my personal portfolio.

Click here to interact with this dashboard on Tableau public