Intro

TikTok is widely used by millions of people, and music plays a big part in the experience. Whether it’s for dance challenges, lip-syncs, or creative videos, music plays a key role in how users engage with the app. Certain songs often gain popularity due to their energy and danceability. Understanding why these songs resonate with users can provide insight into the elements that make them go viral.

The dataset used for this project was taken from Kaggle, specifically the Tiktok Trending Tracks dataset. There are 6746 observations and 23 variables in this dataset. Out of the 23 variables, we used five variables and 3367 observations. Within the five variables, there are two categorical variables and three quantitative variables. The categorical variables are track name and artist name. The quantitative variables are pop_grouped (popularity from 0 of 100), avg_danceability (danceability from 0 to 1), and avg_energy (energy from 0 to 1).

Our leading questions for this dataset were as follows: Does a song’s release date affect its popularity on TikTok? Do older songs have lower popularity, while newer releases tend to be more popular?

Plot one

The first graph we made reflected the Top 10 most popular songs on TikTok and their average popularity score (scale of 95 or higher) out of 100. The x-scale is intentionally zoomed into the range of 90-100 for closer observation. Each bar represents a song that has gone viral on the platform (the dataset was last updated in 2021). Considering TikTok is very music-oriented, it makes sense that many of the songs used come from a variety of genres, including tracks frequently used in challenges and dance trends. An example is “MONTERO (Call Me By Your Name) by Lil Nas X, who, thanks to the corresponding dance, reached a score of 95.”Astronaut in the Ocean,” another hit pop song with a score of 96, shows how TikTok opens users up to genres they may not normally listen to.

Plot two

Next, we explored the relationship between a song’s release date and its average popularity. This graph has a moderate positive correlation. It is left-skewed, linear, and has no outliers, showing that new songs tend to interest audiences. The closer a song is to the present, the higher its popularity tends to be, as reflected in the correlation coefficient of 0.726. For example, Justin Bieber’s Peaches (feat. Daniel Caesar & Giveon), released in 2021, has a popularity score of 99. In contrast, Drake’s Passionfruit, released in 2019, has a popularity score of 83. This trend suggests that listeners gravitate toward newer releases.

Plot three

Finally, we explored the relationship between duration and release date. In this scattergram, a moderate positive correlation can be seen as supported by a correlation coefficient of 0.75. This statistic suggests that the duration of songs got shorter over time. Additionally, it has a heavier concentration of songs released in recent years. It should be noted that the oldest release date is from 1977, and the latest is June 2021. There are a couple of outliers, including the one on the top left of the graph. It is Mirrors by Justin Timberlake, which has a duration of 8.06 minutes. Despite this significant outlier, the data reflects a generally steady trend.

Linear Regression Analysis

A linear regression was conducted to evaluate the relationship between a song’s release date and its popularity on TikTok. The model revealed a statistically significant relationship with p < 0.001, indicating that a song’s release date is associated with its popularity. The R-squared value of 0.527 suggests that approximately 52.7% of the variability in popularity scores can be explained by release date. The null hypothesis, which states that a song’s release date has no effect on its popularity, is therefore rejected. Instead, we accept the alternative hypothesis: newer songs tend to have significantly higher popularity on TikTok than older songs. Furthermore, several release dates had notable negative coefficients around -2, indicating that songs released on those dates had lower popularity compared to the reference date. This suggests that while newer songs are generally more popular, some dates have distinct dips in popularity.

Conclusion

Overall, the visualizations and regression analysis suggest that newer songs tend to be more popular. Visualization 1 highlights the top 10 most popular tracks, while visualization 2 demonstrates a positive correlation between release date and popularity. Visualization 3 shows that more recent songs often have shorter durations, further supporting a relationship between release timing, song length, and popularity. In conclusion, the data suggests that newer releases and shorter durations are linked with higher average popularity on TikTok.