Seattle Library Checkouts

This project aims to develop predictive insights into library book checkouts using machine learning and time series analysis. Leveraging the “Checkouts by Title” dataset provided by the Seattle Public Library, our analysis addresses two main questions: (1) can we predict the number of checkouts in the first year for a new book based on a number of features, and (2) can we forecast future checkouts over several months by analyzing past checkout patterns? Some methods we use include: linear regression, lasso regression, random forest, extra trees, k-nearest neighbors, XG boost, neural networks, STL forecasting, and AutoARIMA.

You can find a detailed report, results, and code here.