Predicting Book Popularity on Goodreads: A Comparative Study of Linear Regression, Decision Tree, and Random Forest

Abstract

The increasing volume of digital book data on social reading platforms such as Goodreads has created an opportunity to analyze factors that influence book popularity using machine learning methods. This study aims to identify which features significantly affect book popularity scores and to compare the predictive performance of three machine learning models: Linear Regression, Decision Tree, and Random Forest. The dataset used consists of 1,296 book records from the Goodreads favorite book list, comprising variables such as has_series, series_number, avg_rating, total_rating, and voters. After preprocessing including the removal of missing values and duplicate data—1,191 records were retained for modeling. The models were evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R²). Results indicate that the Random Forest model achieved the best performance with R² = 0.9722, MAE = 15.82, and RMSE = 88.13, outperforming both Decision Tree (R² = 0.9676) and Linear Regression (R² = 0.8479). Feature importance analysis revealed that the voters variable has the most dominant influence on book popularity with an importance score of 0.9794. This study contributes to the understanding of digital reading behavior and provides a foundation for developing more accurate book recommendation systems.