On April 27th, the China Mobile Wutong Cup National Finals took place at Xiamen University. This competition serves as a platform to foster research and innovation among university students, spanning various fields including artificial intelligence and data security. It emphasizes the integration of big data applications across diverse industries.
At the finals, 15 top teams from across the nation engaged in a pinnacle competition showcasing innovative thinking and technical prowess.
During the China Mobile Wutong Cup National Finals at Xiamen University on April 27th, Mr. Zhao Qingsen from the Department of Applied Biology and Chemical Technology (ABCT) and his team demonstrated a meticulous approach to their project:
During the model-building process, they prioritized data integrity and feature correlation from the outset. The team recognized the importance of the data preprocessing stage, conducted a detailed analysis of missing values in the dataset, and developed a filling strategy based on feature importance and correlation. Their methods included using techniques such as global means and group medians to fill in missing values. When selecting machine learning models, they made customized choices based on competition requirements and business needs. Their toolkit encompassed various algorithms, including adding time series dimensions, convolutional feature extraction, and k-means clustering analysis. To enhance model performance, they implemented a stacking strategy.
Throughout the project, the team remained focused on the model's generalization ability and stability. They carefully optimized model hyperparameters through various tuning methods. Notably, they used techniques such as grid search to fine-tune key parameters like n_estimators and learning_rate, ensuring the predictability and consistency of prediction results.
Mr. Zhao Qingsen's team's achievements in the competition serve as a testament to their unwavering pursuit of excellence and innovative spirit in academic and research pursuits.
