Recently, the intelligent prediction team established by China Post Information Technology (Beijing) Company Limit participated in a competition on Kaggle, and won the silver medal in a competition of Research featuring scientific and scholarly challenges, ranking the 62nd among 2,893 teams, top 2 percent worldwide. With their excellent performance, the team has been certified as a Competitions Master. Among the 190,000 teams that have participated in all Competitions since the establishment of Kaggle, this team ranks 546th, the top 0.29 percent worldwide.
As is known, Kaggle Competitions platform, founded in 2010, is the world's largest data science and machine learning community, and a data science competition platform with the highest recognition in the world. The predictive algorithms used by the intelligent prediction team was independently developed by the Data Application Department of the company, and has been applied in many postal service scenarios, such as predicting the mail volume of collection, delivery and processing. Among them, the model prediction accuracy of the flux and flow direction within the postal network during the "Double 11" peak season is 3.89 percent higher than that of the rival’s. The prediction accuracy of the collection and delivery volume reached 96.3 percent, 1.3 percent higher than the rival’s public data. With the average accuracy reaching 96.6 percent, mail center volume prediction has been applied to scenarios such as planning work shifts, scheduling vehicles, organizing and managing peak season production and so on.