Bagelcode, User Behavior Prediction Paper Accepted at 'KDD 2023'

Bagelcode announced on the 25th that its paper on ‘Mobile Game User Behavior Analysis and Prediction’ was accepted at ‘KDD 2023’.

KDD is the premier international conference in the fields of data mining and artificial intelligence. The annual KDD conference focuses on science and engineering, facilitating active technical exchange and discussion among researchers and industry experts in data science, data mining, large-scale data analysis, and big data.

The paper from Bagelcode’s Data & AI team accepted at ‘KDD 2023’ centers on expanding the scope of data utilization and developing models to acquire relevant information, focusing on mobile game user lifetime value (LTV), purchase prediction, and behavior analysis.

Selected for the oral session, which is rated in the top 1% of papers submitted to ‘KDD 2023,’ the team will attend and present at the conference held in Long Beach, California, USA, starting August 6.

The Bagelcode Data & AI team developed the ‘Multi-Data Source LTV User Representation Model’ (MDLUR), which increases data utilization while addressing the ‘curse of dimensionality’?where accuracy decreases as data usage increases?thereby enhancing efficiency and accuracy even in the analysis of vast datasets. This model demonstrated a 20% performance improvement over existing models in LTV prediction and is effectively used in various data prediction tasks such as payment, conversion, churn, and retention.

Juhyun Kim, General Director of Bagelcode Data & AI, said, “We are pleased that Bagelcode’s Data & AI capabilities have been recognized, and this achievement is the result of active investment in the related field since the company’s inception.” He added, “This year, we aim to apply MDLUR to the casual games currently under development to achieve successful market entry and operational efficiency.”