Kai Yang, PhD student from the Department of Information Systems, together with two undergraduate students from South China University of Technology, triumphed in the JD Discovery-2018 Global Digitalization Challenge (Hong Kong division) and entered a 24-hour offline limit global challenge held in Beijing on 12–13 January. The team was eventually awarded Second Prize in the Global Final, along with RMB200,000.
Organised by JD Digits, a leading digital technology company in China, JD Discovery-2018 Global Digitalization Challenge aimed to connect technical talents with digital elites. It was a real application global scenario design competition. AI and big data experts, industry leaders, and top investors joined the competition as mentors and judges.
Led by Kai, the team outperformed 75 teams and won the Hong Kong championship division in December 2018, after rigorous review by the competition's expert committee.
On 12–13 January 2019, the team joined a final round and competed with other winning teams from the Mainland China, USA and Israel divisions in Beijing. Thanks to their top scoring business plan, they were eventually awarded Second Prize in the Global Final.
Kai was excited about the team’s achievement, and he shared that the biggest challenge in the competition was the unsatisfying performance of ARIMA, an autoregression algorithm model for forecasting. It failed to capture the patterns of non-periodic changes during data collection. “To overcome this challenge, we tried to improve our model by adding other algorithms. We then tested their performance and found that a tree-based learning algorithm, namely LightGBM, performed the best,” Kai elaborated.
