Book sales data is an important component in supporting marketing strategies and managerial decision-making. The objective of this research is to evaluate and compare the effectiveness of the C4.5 and Naive Bayes in processing book sales data at PT. Sonpedia Publishing Indonesia. The dataset used consists of 299 book sales records, processed using RapidMiner software with two validation methods, namely Split Data (80:20) and 10-fold cross validation. Experimental results reveal that the C4.5 algorithm with the split data method obtained an accuracy 88.33%, precision 94.29%, recall 86.84%, and F-Score 90.41%. Using 10-Fold Cross Validation, the performance decreased with an accuracy 86.60%, precision of 92.53%, recall 85.64%, and F-Score 88,99%. In contrast, the Naïve Bayes algorithm demonstrated better and consistent performance. With the Split Data method (80:20), it obtained an accuracy 90.00%, precision 90.00%, recall 94.74%, and an F-Score 92.31%. Furthermore, its performance improved with 10-Fold Cross Validation, achieving an accuracy 91.29%, precision 92.63%, recall 93.62%, and F1-Score of 93.10%. These findings suggest that naive bayes produces more consistent and accurate classification results compared to C4.5. The research is intended to act as a guide for the development of book sales prediction systems that support the effetiveness and efficiency of bussiness decision making.