Multi-Level Deep Cascade Trees for Conversion Rate Prediction in Recommendation System

Developing effective and efficient recommendation methods is very challenging for modern e-commerce platforms. Generally speaking, two essential modules named "Click-Through Rate Prediction" (textitCTR) and "Conversion Rate Prediction" (textitCVR) are included, where textitCVR module is a crucial factor that affects the final purchasing volume directly. However, it is indeed very challenging due to its sparseness nature. In this paper, we tackle this problem by proposing multi-Level Deep Cascade Trees (textitldcTree), which is a novel decision tree ensemble approach. It leverages deep cascade structures by stacking Gradient Boosting Decision Trees (textitGBDT) to effectively learn feature representation. In addition, we propose to utilize the cross-entropy in each tree of the preceding textitGBDT as the input feature representation for next level textitGBDT, which has a clear explanation, i.e., a traversal from root to leaf nodes in the next level textitGBDT corresponds to the combination of certain traversals in the preceding textitGBDT. The deep cascade structure and the combination rule enable the proposed textitldcTree to have a stronger distributed feature representation ability. Moreover, inspired by ensemble learning, we propose an Ensemble textitldcTree (textitE-ldcTree) to encourage the model's diversity and enhance the representation ability further. Finally, we propose an improved Feature learning method based on textitEldcTree (textitF-EldcTree) for taking adequate use of weak and strong correlation features identified by pre-trained textitGBDT models. Experimental results on off-line data set and online deployment demonstrate the effectiveness of the proposed methods.