About the Short Course 

Valid inference results rely not only on sensible modelling and analysis methods, but also on good quality data. Noisy data with measurement error or missing values, however, are ubiquitous in applications. Measurement error and missing values have been a longstanding concern in various fields, including clinical trials, cancer research, epidemiological studies, environmental studies, economics, survey sampling, etc. Investigators often confront noisy data with the urgent need to address the effects of measurement error or missing values in their data analysis. Even though intensive research has been directed to measurement error problems, measurement error methods have not been used frequently in situations that merit their use. Lack of adequate understanding of measurement error effects and necessary analytical skills for handling them are perhaps the primary reasons.

The objective of this course is to lead the audience to visit these challenging areas. Specifically, the impact of measurement error and missing data will be demonstrated, and different measurement error models and missing data mechanisms will be described. Typical inference strategies for handling noisy data will be introduced. Some methods in the machine learning community will also be briefly visited. The discussions will be illustrated with examples and applications.

In-Person Event. Location Of Short Courses: University of Ottawa

Who is this course for?

This short course is designed to serve as a quick tour for the audience to gain insight into sensible analyses of noisy data. The anticipated audience includes applied statisticians, graduate students, researchers, and data scientists who are interested in having an overview of measurement error and missing data.

Level Of Instruction: Beginner

Learning Outcomes

The objectives of this course are to:
- alert the audience to the impact of measurement error and missing data on data analysis;
- provide the audience an introduction to measurement error models and an overview of the strategies of handling missing data problems;
- bring together assorted methods under the same umbrella and deliver them at an accessible level of applied statisticians, graduate students, and data scientists.

Course Materials

This short course will cover two primary parts together with a final part which will briefly discuss emerging issues. The first part includes various issues of handling measurement error problems, and the second part deals with different topics concerning missing data.

Part 1: Handling measurement error

1. Introduction, measurement error examples
2. Discussions of measurement error effects
3. Measurement Error Mechanisms and Models
4. Inference methods of accommodating measurement error effects
   • Regression calibration
   • Simulation extrapolation

Part 2: Handling missing data

1. Introduction, missing data examples
2. Missing data mechanisms
3. Analysis method
   • Imputation method
   • Likelihood-based method
   • Inverse Probability Weighted GEE

Part 3: Discussion of emerging issues

1. What if both measurement error and missing data co-exist.
2. What if there are other features.
3. Some work related to the machine learning techniques

Delivery Structure

The course will be delivered by the instructor in the lecture format, with the Q&A periods provided to engage the audience.

Knowledge Assumed

Having basic statistics knowledge such as regression analysis would be useful to well appreciate this course.

About the instructor: Grace Y. Yi

Grace Y. Yi is a professor in the Department of Statistical and Actuarial Sciences and the Department of Computer Science at the University of Western Ontario. She currently holds a Tier I Canada Research Chair in Data Science. Dr. Yi's research interests focus on developing statistical methodology to address challenges concerning measurement error, causal inference, imaging data, missing data, high dimensional data, survival data, and longitudinal data. She authored the monograph “Statistical Analysis with Measurement Error or Misclassification: Strategy, Method and Application” (2017, Springer), and coedited “Handbook of Measurement Error Models” (with Aurore Delaigle and Paul Gustafson) (2021, Chapman & Hall/CRC). Dr. Yi received her Ph.D. in Statistics from the University of Toronto in 2000 and then joined the University of Waterloo as a postdoctoral fellow (2000-2001), Assistant Professor (2001-2004), Associate Professor (2004-2010), Professor (2010-2019), and University Research Chair (2011-2018). She is a Fellow of the Institute of Mathematical Statistics, a Fellow of the American Statistical Association, and an Elected Member of the International Statistical Institute. In 2010 Dr. Yi received the Centre de Recherches Mathmatiques and the Statistical Society of Canada (CRM-SSC) Prize which recognizes a statistical scientist's excellence and accomplishments in research during the first fifteen years after earning their doctorate. She was a recipient of the University Faculty Award (2004-2009) granted by the Natural Sciences and Engineering Research Council of Canada. Dr. Yi has served the profession in various capacities.

She is currently a Co-Editor-in-Chief of The Electronic Journal of Statistics (2022-2024) and the Editor of the Statistical Methodology section for The New England Journal of Statistics in Data Science. She was the Editor-in-Chief of The Canadian Journal of Statistics (2016-2018). Dr. Yi served the Statistical Society of Canada as President-Elect, President and Past President for 2020-2023. She was President of the Biostatistics Section of the Statistical Society of Canada in 2016. She is currently the Chair of the Lifetime Data Science Section of the American Statistical Association. She is the Founder of the first chapter (Canada Chapter, established in 2012) of the International Chinese Statistical Association.

Affiliations: University of Western Ontario