Once upon a time, I read this cool X-post shown below. Yan here is trying to create a story from the Craiglist’s rental titles in the San Francisco Bay Area. The graph is a bit advanced because it is using density distributions and a lot of statistics… However, he provides a nice Legend explaining how to interpret the data and the visuals help to get the main message: “Nice and Clean are used for low-rent ads?”.

All of a sudden I started to wonder what would happen if we do a similar analysis with the job offers and their salary! Evil, right? 🙂 The result is shown in Figure 1.

I downloaded the dataset from Kaggle “jobs-in-data” and did some text cleaning with a lot of copy-paste from this page (in case you are interested). Then, I refactored a lot Yan’s code to make it work with my data. The result can be summarised with a poem:

In the world of job descriptions, there’s a divide,
‘Strategic’ and ‘administrative’ are often underpaid,
Their work is varied, but their pay doesn’t provide,
It’s tough for them, their efforts often dismayed.

But look over there, a different scene is found,
‘Good’ and ‘corporate’ roles, they get paid quite well,
Their performance is measured, their pay does astound,
In this world, they’re the ones who excel.

So here’s to those with various tasks, they bear,
May they find joy in their work, no matter what’s told,
And to the well-paid, with success in the air,
May they continue to thrive and their fortunes unfold.

In job descriptions’ world, this tale is quite bold,
Where worth and wage, in verse, are artfully told.

ChatGPT Poem

My evil DS is now wondering… what happens if we do this same analysis but for Glassdoor company scores instead of Salary offered? We would only need to join this data with the one used here already. Let me share my notebook with the code if you accept the challenge!