September 2021 - December 2021 (3 months)
This is a musical case study of the public sentiment using the lyrics of the top 20 songs of each month ranked by the YouTube Music Charts. Through this project, we hope to understand whether popular music trends could be a useful indicator of public sentiments in a global event.
I wanted to use my understanding of data visualization principles to create an application that gives me insights into a topic of my interests while practicing interactive visualization authoring with Tableau.
Data cleaning/analysis
Visualization design
Inspired by personalized music playlist on Spotify, our data adventure starts with a wonder: What does music listening trends say about the listeners' emotions?
But there is so much to music -- when we talk about music we can talk about its acoustic features like the key, tempo, and pitch; or we could talk about its instrumentation and how the combination of strings make it sound epic; or we could talk about the lyrics.
We performed an extensive background research on past and current studies on music sentiments. The majority of them focus on the acoustic properties of music and confirms that there exists correlation between the acoustic of music and the listener's sentiment. These findings spark our interests: what about the lyrics?
So we rewrote our question: Can we use lyrics from musical trends to analyze public sentiments and how would we visualize it?
We had an hour long brainstorming session on how we could transform the lyrics into numerical data and how we could present it in an informative and engaging way.
The entire project is made up of two Tableau dashboards, the map gives an overview of the sentiment for each of the seven countries, whereas the line & bar charts provide information on the correlation between different sentiments and severity of the pandemic (measured in new case rate).
If you are interested in knowing more about our data collection process and the limitation of our case study, please read our documentation on the design process.
The process of designing data visualization for music was enlightening and inspiring and I have also learned some lessons I could take to future projects.
We were ambitious in the very beginning — we envisioned a world map with data from at least ten countries covering all continents and densely populated countries using highly influential charts from the industry (billboard 100, etc.). However, as summarized in the project report, we encountered limitations in the available dataset and resources. Realizing that we would have to manually compute the sentiment data from all lyrics (including non-English lyrics) and would need to pay to access major chart archives, my partner and I made the decision of narrowing the scope of our project to include only 7 countries and analyzing only top 20 songs of each month.
With this decision, we were able to maintain a good balance between the size and the quality of our project, delivering a product that is both rich in content and engaging in experience. I will use this project as an example to guide my future work to not be overly ambitious in the initial stage of product design and always be realistic about the limitations I face in the process.
One of the design challenges we encountered was picking colors for the three sentiment values — positive, neutral, and negative. We first went with the palette of blue, green, red for the aesthetics; however, during our informal usability testing, we found that our color choices were not intuitive in their meaning. There was some confusion about the association between colors and sentiments, specifically, some users view green as positive and blue as neutral, which was the opposite of what we assigned. After rounds of iteration experimenting with different shades and combinations, we settled on our current choice. These particular shades of green, yellow, and red (I call this palette “the vegetable mix”) are distinguishable across all opacity and visually pleasing.
This process has taught me that design often time requires some exploration between aesthetics and usability, and we really need to pay attention to the cultural and societal connotations of our design choices to avoid unnecessary complications in understanding and using our design.