Efficient Click-Through Rate Prediction for Developing Countries via Tabular Learning

TL;DR

This paper demonstrates that tabular learning models are more efficient and effective for CTR prediction in developing countries, outperforming traditional over-parameterized models in accuracy and resource usage, and improving real-world user engagement.

Contribution

It introduces the application of tabular learning models to CTR prediction, showing they outperform existing models in accuracy and efficiency, especially in resource-constrained environments.

Findings

Tabular learning models outperform state-of-the-art CTR models on multiple datasets.

Tabular models require less computational resources and can be trained faster.

Online A/B testing confirms improved CTR with tabular learning models.

Abstract

Despite the rapid growth of online advertisement in developing countries, existing highly over-parameterized Click-Through Rate (CTR) prediction models are difficult to be deployed due to the limited computing resources. In this paper, by bridging the relationship between CTR prediction task and tabular learning, we present that tabular learning models are more efficient and effective in CTR prediction than over-parameterized CTR prediction models. Extensive experiments on eight public CTR prediction datasets show that tabular learning models outperform twelve state-of-the-art CTR prediction models. Furthermore, compared to over-parameterized CTR prediction models, tabular learning models can be fast trained without expensive computing resources including high-performance GPUs. Finally, through an A/B test on an actual online application, we show that tabular learning models improve not only offline performance but also the CTR of real users.