Key Takeaways from the 2025 China University Big Data Challenge

Team Overview

Team Name: 虎鲸qwq

Member: 刘航 (Zhejiang Ocean University)

National Rank: 8th place

Competition Context

The 2025 China University Computer Competition – Big Data Challenge required participants to predict the price movement rankings of Shanghai‑Shenzhen 300 index component stocks. The task emphasized handling structured time‑series data and forecasting extreme value rankings.

Personal Journey and Learning

As a first‑time finalist, the author reflects on transitioning from theoretical knowledge to practical implementation. The competition served as a bridge between academic concepts and real‑world data engineering, highlighting the importance of problem decomposition, solution implementation, and iterative optimization.

Core Technical Challenges

Designing models capable of predicting extreme stock price changes.

Managing noisy and irregular data formats.

Balancing model complexity with generalization ability.

Addressing these challenges required a solid theoretical foundation, flexible adaptation to data characteristics, and systematic experimentation.

Four Key Takeaways

Align Solutions with Data Logic: Tailor approaches to the specific structure and nuances of the dataset to find effective pathways.

Iterative Optimization Over One‑Shot Solutions: Break down tasks into manageable steps, prioritize components, and refine progressively for better cost‑benefit outcomes.

Persistence Pays Off: Overcoming bottlenecks often involves consulting additional resources or re‑running experiments, which fuels growth and discovery.

Collaborative Insight Enhances Models: Diverse perspectives on feature selection and model improvement lead to more robust solutions.

Future Outlook

The experience reinforced that big data is a practical tool for solving real problems, not just abstract numbers. The developed framework—focusing on time‑dependency and extreme‑value prediction—can be adapted to domains such as cryptocurrency, commodity futures, and retail sales forecasting.

Acknowledgements

The author thanks the competition organizers, Tsinghua University for support, the judges for insightful questions, and personal perseverance during countless late‑night debugging sessions.

Illustrative Slides