The effect of release dates on the book sale ranks

Official URL: https://risc01.sabanciuniv.edu/record=b2549543_(Table of contents)

In this study, we examined the effect of a book’s publishing date on its sales ranking with a Linear Regression model by using Amazon’s daily book ranking and price data for 67 days. We found out that the release date of a book is an important factor in evaluating the book rankings. We also studied the prediction of rankings using the lagged variables of price and ranking. To transform this time series prediction problem into a supervised learning problem, we used the sliding window approach. We used four machine learning and one deep learning approach to predict the rankings. To compare the results, two evaluation criterias; R2 and root mean squared error were used. When tuning the hyperparameters, we used k-fold Cross Validation. We found out that linear regression outperformed the rest of the models, which are Ridge Regression, Random Forest, Light Gradient Boosting Machine, and Neural Network