Feedback from Simon Frew

[srfrew]

Great work overall! This dashboard visualizes key information in the music industry, and would definitely allow for record companies to make informed decisions about artists and genres. The business proposition here is very useful and not one I've heard of - usually these kinds of music visualizations are more for personal use. The filters and use cases in the proposal and dashboard would be very useful! The feedback below is in no particular order, but included use case questions, UI, user flow, some definitions and charting questions. Hopefully this helps and best of luck in the next milestone!

• It may be useful to record labels to view the consistency of performance over time for an individual artist. Highlighting the distribution of an artist's discography may highlight whether they are consistently high performing or only have a few hits. This may be more useful than a top-song list, and allow for better comparison between artists!
• Seeing as the proposal is attempting to help record labels track individual artists, a summary card for each artist selected may be very useful (e.g., their style, number of songs, proportion explicit, etc). Many of these aspects are included in the dashboard already but in aggregate across artists. Breaking this up and organizing by artist may allow for more better comparisons!
• Popularity of top songs: unclear what the shading means (speechiness or explicit), consider adding a label, legend, or tooltip to clarify
• Popularity by Speechiness over Time: unclear what "speechiness" means, and the items I selected largely had low speechiness. I assumed that it was showing by all artists, but I struggle to understand the graph.
• In fullscreen, the right sidebar margin takes up a large amount of space for static text. Consider reducing this so that the charts themselves could be larger, and potentially aligning the "about" section at the bottom.
• The mean popularity of songs by type and artist chart is insightful. I would prefer that it is transposed so the categories are on the "Y" axis, and popularity on the X axis. This will make it easier to read the labels and compare numerically between artists. Additionally, it may not be necessary to have both Clean and Explicit labels repeating as they are now. The color legend itself should suffice.
• The standard deviation of the top 5 songs may be useful to include, as if widely different artists are measured, there may be issues with accurately communicating the distribution. (e.g., if the top 5 were bimodally distributed.
• I noticed that the "genre" selector resets the dropdown contents for "artist", however, this does not clear existing artists from the selection. This may lead to confusing interactions, for example, I was able to select "Shakira" from Latin then move to Dance Pop and select the Jonas Brothers. I am unsure if this is intended though.
• This may be very minor: Some of the terms in the Spotify dataset (valence, speechiness) may not be well-known among record labels due to it being a Spotify-specific designation. It may be worth providing a tooltip with definitions in case the user is confused.