This project is a Python-based application that focuses on predicting the listening time of each artist by users in the next week for a music streaming service called "Pozytywka". The project uses two experimental models (Model A and Model B) dummy moving average and SARIMA.
Our service earns when people listen to music. If we don't know how much time they will listen to it in the future, it's hard for us to settle with artists and negotiate rates with them. The business task is to predict the listening time of each artist by users in the next week. The business success criterion is higher accuracy than the static approach - the average of the previous four weeks.
- Python
- Pandas
- statsmodels
- autoArima
- FastAPI