Time Series Analysis and Forecasting with Spotify Song Streams.
The article dives into Spotify’s streaming data for the band called Among Radio Thieves. A rock band that happens to have the author of the article as a bass player.
The data-driven exploration unveils streaming patterns, utilizes model selection techniques, and in the end forecasts song streams for the upcoming three months.
The analysis starts with the collection of data in csv format from Spotify. The dataset contains monthly streams from their releases in 2019 to the current year.
In the first year, streams show a downward trend before stabilizing between 20 and 40 streams per month. Furthermore, it appears the streams show seasonality with three peaks per year. In the first two years, these peaks occur in winter, the fourth, and the eighth month. The following three years show a small variation, with peaks in both winter and the fourth and sixth months.
To ensure robust modeling, the Augmented Dickey-Fuller test is used. With a test statistic of -7.40 and an extremely low p-value (7.47e-11), confidently can be confirmed that the data are stationary.
0 Comments