Conversion Rate Prediction

Conversion rate prediction (CVR prediction) aims to accurately estimate the likelihood of a user completing a desired action, such as a purchase or sign-up, crucial for optimizing online advertising and recommender systems. Recent research emphasizes addressing data sparsity and distribution shifts, employing techniques like self-supervised pre-training, contrastive learning, and doubly robust methods to improve model accuracy and calibration, often within deep learning frameworks such as neural networks and factorization machines. These advancements are vital for enhancing the effectiveness of online platforms by improving targeting, resource allocation, and ultimately, revenue generation.