Recently, Jie Li, a 2017 fifth-year Phd student and Hu Qirui, a 2020 second-year Phd student in our center, won the 2021 ISI Jan Tinbergen Award Division A First Prize from the International Statistical Institute (ISI).
The International Statistical Institute (ISI), headquartered in the Netherlands, is one of the three authoritative statistical academic organizations in the world. It aims to lead, support and promote the understanding, development and good practice of statistics worldwide. The various honors and awards issued by ISI are highly recognized by the international statistical community. The ISI Jan Tinbergen Awards are an opportunity for young statisticians to present their papers at the biennial World Statistics Congresses (WSC). The awards are named after the famous Dutch econometrician and Nobel Prize winner, and are sponsored by the Dutch ‘Stichting Internationaal Statistisch Studiefonds’ (International Statistical Study Fund Foundation). Papers in Division A are to address an applied statistical problem of real interest in countries with a limited statistical infrastructure.
From 2019, the winners are no longer limited to developing countries. Since 2013, a total of 14 people from various countries have won awards, of which three are Chinese. Li Jie and Hu Qirui are also the first Chinese scholars to appear in the first prize list. Li Jie and Hu Qirui received a prize of 2500 Euros and were invited to participate in the 63rd ISI World Statistics Congress held in The Hague, Netherlands from July 11 to 16 (finally held online due to the epidemic), and made an invited talk at the Jan Tinbergen Awards Session.
Li Jie and Hu Qirui’s award-winning paper "Prediction Interval of Air Pollutants Concentration by Nonparametric Regression Analysis" applied non-parametric regression model to locally stationary time series and analyzed the daily concentration data of six major air pollutants in Xi'an from 2013 to 2020, which was provided by Dr. Zhang Fengying, a senior engineer of the China Environment Monitoring Center.
The paper proposed to use spline regression to estimate the trend function and kernel regression to estimate the variance function. Quantile estimator is obtained after fitting the autoregressive model of errors and prediction interval for multi-step-ahead future observation is constructed using the estimated quantiles. Compared with the prediction interval obtained by traditional methods such as seasonal difference integrated moving average autoregression (Seasonal ARIMA), the prediction interval obtained by the proposed method in the paper is not only narrower in length, but also has better prediction accuracy. The proposed method effectively interprets the underlying dynamics of air pollutants concentration data as well as forecast the future concentration, which is helpful for pollutant management and early prevention. It is also worthy to mention that the award-winning paper of Li Jie and Hu Qirui was completed without the direct participation of instructors.