Creating a Power BI Dashboard for Labour Market Analysis

Analysis of job satisfaction and career switching success among data professionals

Project Summary

Business objective: Given survey results of data professionals, identify their average job satisfaction, compensation and likelihood of successful career change into data.

Outcome: Job satisfaction of data professionals surveyed is at a medium level with the majority of respondents seeking better remuneration packages. Career switching into data shows promise with almost 60% reporting such a move.

A Power BI project is a must for any aspiring data analyst’s portfolio. I hope this article will give you an idea about exactly how to create your first one if you’re thinking of how to get started. After all, this is my very first Power BI dashboard too!

Building on Alex’s project

This work is inspired by the project created by Alex the Analyst on his YouTube channel. Check out his video where he runs you through the creation process.

While Alex demonstrates the basic tools, I’ll be running you through my own design process of the dashboard, and some advanced functions I decided to adopt. I hope this is a fitting add-on to the wonderful work Alex does (Thank you, Alex!).

Please bear in mind that there would undoubtedly be better ways of doing these tasks, and some ideas mentioned herewith would be downright incorrect. If you find any, please leave a comment below. As beginners, we should embrace our mistakes and not procrastinate for the sake of perfection. So here goes version one!

Dataset

The dataset was collected by Alex himself, as he mentions in his video, through a survey he conducted. You can download it from his GitHub repo, the link to which is mentioned in the description of his video.

The survey has been done on data professionals, asking about their careers, education, and salaries. This data is interesting to us as beginners since it is real-world data about the industry we’re aspiring to step into.

Data Cleaning

The dataset requires some cleaning by removing blank and irrelevant columns, renaming headers, and converting salary ranges to mean values. The much harder task would be dealing with ‘other’ responses.

The cleaning process can be done in both Excel and Power BI through Power Query. For example, the splitting of columns to extract ‘other’ responses can be done in Power BI in a similar manner to Excel.

Alex covers the cleaning process through Power Query in Power BI. This is likely to be a new area for you to explore and gain knowledge in.

If done to the complete extent, the toughest column to clean is likely to be the ‘country ’column, which only had four preset answers. Along with many possible other responses it had spelling mistakes as well. I have opted not to extract each country since my sole focus was building the dashboard.

My Design Process

The dataset contains a wealth of information and I thought it would be best to focus on three major areas: overview of the participants, their salaries, and how they broke into the field. To not overwhelm the audience, I have dedicated an individual slide to these focus areas.

To simulate a conversation between the data and the audience, I doubled down on natural questions and asked them in the style of a question itself.

Who were they? The overview of the participants

Before drawing any conclusions, it is imperative to understand the context of the data we’re viewing. The first slide was dedicated to displaying the basic details of the data professionals who participated in this survey.

Students and unemployed respondents were shown separately since their responses are likely to be at odds with the rest. (Maybe you can have a feature to opt them out entirely from the displayed results!)

The list of countries shows that the majority of the responses are from the US. It can be used to filter the entire dashboard if needed to focus on specific labor markets.

The drill-down feature was enabled in the ‘Who took the survey’ section. Focusing on the overwhelming majority of data analysts at first take, it allows deeper analysis of other roles if needed. (With more cleaning, the ‘other’ section can be further expanded.)

How are they paid? Summary of salary information

‘How is the pay?’ is possibly the top question when choosing a career path and this survey did not disappoint!

It is interesting to see that ‘better salary’ is the top current priority of the majority of data professionals. This is reinforced by the low average rating for their salary satisfaction (4.27).

Depending on the conditions of the labor market, the pay scales can widely vary. While the overall average pay for a data analyst is $55,000, if you filter on the US, you can see that it rises to $80,000. All other roles show a similar trend. This understanding is critical when gauging our salary expectations while on the job hunt.

Breaking into the World of Data

The fact that almost 60% of the respondents have switched careers into data fills me with hope! I centered this card to show it as the main takeaway from this slide. If you’re aiming to do the same, I guess we have very good odds, my friend!

I hypothesized that depending on their experience level and roles, people would have had varying difficulties in breaking into the field. To incorporate this, I replicated the roles chart in this slide. As seen below, the students skewed more towards the difficult side (17% of the responded ‘Very Difficult’ to the switch as opposed to the average 7%).

Key Learning Point: Calculated measures on a card

As shown above, a card is usually used to display a key data matric. If used directly on the raw data for the column ‘Did you switch careers into Data?’, it would show the results as follows.

Displaying the percentage itself required calculating a separate measure as:

Percentage for Card = DIVIDE(CALCULATE(COUNT('Table Name'[Col Name]),'Table Name'[Col Name] = "VALUE"),COUNT('Table Name'[Col Name]))

This expression uses the CALCULATE function to filter the “YES” responses, counts that number, and divides it by the total responses. I have found that the CALCULATE function is immensely useful in developing custom values, and requires more practice on my part to master.

Key Learning Point: Custom sorting chart axes

Power BI only allows alphabetical and numeric sorting of axes but this can simply not convey the intended message when a more intuitive order is present in the dataset.

Both the education levels and stages of difficulty follow such an order in their responses. For education, the gradually advancing stages are not alphabetical. The same goes for increasing difficulty categories from ‘Very Easy’ to ‘Very Difficult’.

With the default alphabetical ordering, the graphs show up with ‘high school’ between ‘masters’ and ‘bachelor’ and ‘Very Easy’ right next to ‘Very Difficult’.

And sorting with the number of responses, the result is as below. ‘High School’ is still in the middle and ‘Neither easy nor Difficult’ is at the beginning rather than the middle!

Both of these graphs are not ideal. It is required to sort the categories in the natural order, to display the graphs correctly as below.

This can be done by creating a separate table to set the intended order, incorporating that order into the main data table, and ordering the education and response columns from that particular index.

Note that the same table can be ordered independently from two indices, to order two separate charts, as shown above. This process is clearly explained in the following video.

Color scheme

I usually prefer a minimalistic theme with a lot of white space and went with such a scheme for this dashboard. Using a muted blue pallet, I reserved a vibrant shade for the most distinct piece of information in each slide, like the higher average pay for females as seen below!

Do we need a dataset we’re super passionate about?

I’ve read in many places that we should select an area which we’re really passionate about to do a project. There is no doubt that this is true, but if you’re just getting started, it’s easy to get carried away with looking for the perfect data rather than actually creating something! The idea is to get a grasp on the functionalities and processes involved rather than the creation itself.

This particular dataset is far from complete and comprehensive, but getting to know the status quo of the industry was a very pressing concern for me. Looking at the dataset, I did get valuable answers to some of the questions I had, such as my final goal of becoming a data scientist is in fact financially rewarding.

I am glad that I completed writing this Medium article, the first one I ever wrote! The fear of judgment and criticism keeps me from publishing my work, and this might be the case with you too. If you’ve read this far I wholeheartedly thank you and I hope your time has been valuably spent. Let it assure you that your work is still valuable and will help someone out there. So get creating and publishing!

Click here to connect with me on LinkedIn if you have any questions. I’m delighted to assist!

Please leave a comment with your input and a clap to show your support. Follow me on Medium for more articles on data analysis.