Lizhiqiang

Name: Li Zhiqiang Gender: Male

Education and degree: Doctoral student, Major: Non parametric Statistics, Title: Associate Professor

Education and Main Work Experience 

March 2019 -present  Teacher at the Department of Mathematics, School of Mathematics and Physics, Beijing University of Chemical Technology

June 2001-March 2019 Teacher, Mathematics Department, School of Science, Beijing University of Chemical Technology

July 1993-July 1998 Teacher at Baoding No.15 Middle School

September 2004-July 2007, School of Applied Mathematics and Physics, Beijing University of Technology, majoring in Applied Mathematical Statistics, with a PhD in Science and supervised by Professor Xue Liugen

September 1998-May 2001, School of Science, Northern Jiaotong University, majoring in Applied Stochastic Analysis, with a Master's degree in Science and supervised by Professor Liu Kunhui

September 1994-July 1997, Department of Mathematics, Hebei University, majoring in Applied Mathematics, Bachelor of Science

September 1991-June 1993 Baoding Normal College, majoring in Mathematics

Expert in the evaluation of the National Graduate Mathematical Modeling Competition, expert in the master's evaluation of the Ministry of Education's expert database, member of the Big Data Statistics Branch of the China Association for On site Statistics, and member of the China Society of Industrial and Applied Mathematics. Guided all undergraduate and graduate students in the school to participate in mathematical modeling competitions and won more than thirty second and third prizes. Led two horizontal projects and guided more than ten graduate students in the field of applied mathematical statistics. Published more than thirty academic papers in core journals and SCI journals.

Research direction

Inference of Non parametric and Semi parametric Statistical Models

Missing data statistical analysis

Statistical analysis of longitudinal data models

Bayesian statistical modeling of small sample data

Statistical diagnosis and testing of experimental data

Quantitative structure-activity relationship modeling of chemical experimental data

Algorithm research on statistical models under massive data

Research on Robust Statistical Analysis Algorithm Based on Quantile Regression

Introduction to research content

The research areas mainly involve non parametric statistics, longitudinal data statistical analysis, Bayesian statistical analysis, model statistical inference algorithms under massive data, quantile regression, and diagnostic testing of regression models.

The research areas mainly involve non parametric statistics, longitudinal data statistical analysis, Bayesian statistical analysis, model statistical inference algorithms under massive data, quantile regression, and diagnostic testing of regression models.

·Engaged in research on semi parametric statistics under missing data, and achieved certain research results in the estimation theory of partial linear models. In terms of methodology, a series of methods such as restricted maximum likelihood, Profile least squares, EM algorithm, non parametric smoothing, and cross-sectional kernel estimation were mainly used to conduct detailed and in-depth research on partial linear models and partial linear variable coefficient models with non negligible dropouts under missing data. Currently, we have comprehensively understood the history and current situation of the research field of this project, mastered the statistical inference methods and related algorithms of semi parametric models, variable coefficient models, mixed effects models, and longitudinal data.

·Conduct statistical modeling and quantitative structure-activity relationship research on ultra small sample data, screen important influencing indicator variables, and conduct factor analysis. The proposed two-step variable selection method can effectively overcome the problems of multicollinearity and overfitting in small sample data. The problem of change points encountered in experimental data can be solved through methods such as threshold regression.

·Based on the data obtained from orthogonal experimental design, a longitudinal data mixing effect model is used to model and find the optimal chemical reaction conditions. By using quantile regression estimation methods and diagnostic tests of models to perform robust statistical analysis on data, strong influence points and possible outlier data points can be identified.

·Statistical modeling for massive data can effectively reduce computational burden and improve computational efficiency by utilizing methods such as smooth approximation of functions and divide and conquer algorithms to achieve computer parallel computing.

·A method based on model and estimation formula correction is used to eliminate the influence of measurement errors in the data and find the correct model relationship between indicator variables.

·Using machine learning and statistical modeling methods to characterize some stoichiometric problems, such as data classification, indicator prediction, material uniformity, outlier testing, etc., non parametric modeling and testing methods are used to effectively evaluate experimental conclusions.

The undergraduate and graduate courses taught include，

Probability Theory and Mathematical Statistics,

Time Series Analysis, 

Fundamentals of Information Theory, 

Introduction to Mathematics，

Multivariate statistical analysis，

Applied Mathematical Statistics,

Advanced Mathematical Statistics,