Prediction Model of Book Popularity from Goodreads “To Read” and “Worst” Books

With the growth of technology, recommendation systems have become a staple tool for many hobbies. In book communities, book recommendation systems play an important role in recommending a reader their next possible best read. To create a good book recommendation system, it is important to find a compatible machine-learning technique to be able to create accurate predictions for the recommendations. In our research, we focus not only on the book and the user's ratings of the book, but we incorporate the author's ratings as it is an important point of consideration when readers choose their next book. We organically scraped our data set from Goodreads, the current largest online book community. We utilized different machine-learning techniques on our data developed to compare the compatibility and applicability of each based on the algorithms' accuracy. We tested Decision Tree, Random Forest, SVM, Logistic Regression, K-Nearest Neighbor, and XGBoost. We found that XGBoost is the most compatible technique for our data set, with an accuracy of 0.96.