<![CDATA[The rapid development of information technology has transformed consumer shopping behavior, particularly through e-commerce platforms. Online shopping has become a primary trend due to its convenience and the growing penetration of the internet. Understanding online purchase intention is therefore crucial for businesses in devising effective marketing strategies. Purchase intention is influenced by factors such as product quality, price, customer reviews, and platform usability. However, predicting purchase intention poses a significant challenge due to the large and complex nature of consumer data. Smote used for imbalance data. This study aims to combine CNN (Conv1D) and BiLSTM for high-accuracy purchase intention prediction. The research focuses on analyzing model accuracy and the effectiveness of the algorithms in handling imbalanced data. The results indicate that the combined CNN(Conv1D) + BiLSTM model achieves 97% accuracy with balanced evaluation metrics, although the True class recall (96%) is slightly lower than that of the False class (95%). Further optimization is needed to enhance overall model performance.]]>
Copyrights © 2025
