Authors: Peijie Sun, Yifan Wang, Min Zhang, Chuhan Wu, Yan Fang, Hong Zhu, Yuan Fang, Meng Wang

Published on: April 12, 2024

Impact Score: 7.6

Arxiv code: Arxiv:2404.08301

Summary

• What is new: A robust model that predicts user game spending without user IDs, ensuring privacy and featuring a unique approach in representing user preferences and game features.
• Why this is important: The challenge of accurately predicting user spending in mobile gaming due to unpredictable user behavior.
• What the research proposes: A stable model training and evaluation framework that standardizes spending data and a collaborative-enhanced model that separately represents user preferences and game features.
• Results: Achieved a 17.11% improvement on offline data and a 50.65% boost in online A/B testing compared to production models.

Technical Details

Technological frameworks used: model training and evaluation framework

Models used: collaborative-enhanced model

Data used: user spending data

Potential Impact

Mobile gaming industries and companies focusing on maximizing revenue through in-app purchases could benefit significantly.

Want to implement this idea in a business?

We have generated a startup concept here: PredictPlay AI.