Effective hyperparameters tuning is essential to optimize the performance of machine learning models. This study provides a comprehensive review of hyperparameters tuning techniques and explores its theoretical foundations, including grid search (GS), random search (RS), and Bayesian optimization (BO). It then applies these methods to support vector machines (SVM) and K-nearest neighbors (K-NN) in an e-commerce context, specifically focusing on customer churn prediction. The results showed that BO offers a balance between execution time and accuracy, while GS and RS, despite their longer durations, significantly improve model accuracy.