Since December 2022, I’ve wrapped up every month with a playlist containing my current favorites. I liked the idea of being able to look back and relive certain time periods of my life, or see how I was feeling. Coming up on year three, I wanted to see if there were more patterns to be found.

Using Claude to help collate and analyze the data, this is what 38 months of saved Spotify tracks say about my taste and mood.

Spotify’s API doesn’t provide song information like tempo, energy, valence, or “danceability” anymore, so I pulled genre tags from Last.fm’s, and then supplemented that with sentiment analysis on each song’s lyrics.

I didn’t expect the data to map kinda cleanly onto my life. The month I graduated had some of the saddest lyrics I’d saved, and each trip home to Bangkok shows up as a spike in positive sentiment. And upon moving countries, I unconciously replaced a lot of my artist roster.

Here’s my monthly listening volume with the average lyrical sentiment overlaid:

In May 2024, I graduated high school. It’s the saddest month in 38 months of data. I only saved 9 tracks, and with a sentiment score of -0.132, it was the only time this ever went negative.

Something about endings being melancholic?

What genres I listen to

My favorites from the last three years are (very roughly) 40% indie, 32% pop, and then a mix of everything else. Let’s see how these have varied over time:

A few things stand out.

• Indie (green) was most dominant in late 2022 (73% of what I saved that December)
• Pop surges during album cycles: November 2023 with Taylor Swift’s 1989 tv, June 2024 with brat, and August 2025 with Sabrina Carpenter’s Man’s Best Friend.
• Hip-hop was present early on but has almost completely faded from my favorites.

Who I dropped

In August 2024, I moved from Bangkok to Vancouver. That month, I favorited only 7 tracks, but 93% were from artists I had never saved before.

Interestingly, moving month had the highest sentiment of any phase with 0.602. I think maybe it was the optimism of moving somewhere new.

Taylor Swift, who’d been in my top 3 for a year and a half, did not make the trip, while Sabrina Carpenter and beabadoobee, who’d been in the background, became my new stars.

Who stuck around

This gorgeous chart maps when each of my top artists appeared in my favorites. A bigger dot means more tracks from that artist that month. Some artists defined a single era, and others persisted across the whole timeline.

Sabrina Carpenter appeared in 15 different months across 3 years. She’s definitely stuck around, and so has beabadoobee.

What were they singing about?

The sentiment analysis I did on the lyrics of the tracks tagged five recurring themes. Love dominates (it’s pop i guess ), but the smaller themes reveal more.

• Perhaps unsurprisingly, every February seems pretty loaded: in February 2024, I favorited 32 love-tagged tracks and 15 heartbreak tracks.
• Nostalgia seems to peak during transition periods. It spiked in Feb 2024 (last semester of first year), Aug 2025 (leaving home again), and Dec 2022 when I first started tracking.

Going home

This is perhaps the cleanest pattern in the entire dataset: every time I go back home to Bangkok, the music I listen to is happier.

For context, my overall average is 0.46. December 2025 (two weeks at home for New Year’s) hit 0.800, the highest of the entire timeline. Every trip back causes a spike relative to the surrounding months.

Was I surprised?

It was cool that the data made legible something I’d only felt — that moving (really, my environment) changed how I listen. The artists I brought with me, the ones I left behind, what I felt in that graduation month, the brightness of coming home.

I’d really like to keep learning more about effective interactive visualizations. You have to check out pudding.cool/.

Methodology: Data sourced from Spotify saved tracks (Dec 2022 – Jan 2026). Genre classification using Last.fm API tags. Lyrical sentiment analysis using VADER (compound score) on lyrics pulled from Genius. Theme tagging via keyword/NLP classification on lyrics. Analysis done in Python. Visualization in React + Recharts.