Workshops

Incorporating information from several sources is a fundamental problem in statistics. In statistical agencies, the desire to combine information from different sources to obtain an improved official statistic is increasing and survey data integration becomes an emerging area of research. In this workshop, we review the current state-of-the-art methods for survey data integration and discuss some future research direction. Topics includes basic theory of missing data, mass imputation, propensity score weighting for data integration, and nonignorable missing data modeling and estimation.

Jae-Kwang Kim is a Professor of Statistics at Iowa State University and also holds a position at Department of Mathematical Science in KAIST, Korea. He is a fellow of ASA and the recipient of 2015 Gertrude M. Cox award. He is a coauthor of the book Statistical Methods for Handling Incomplete Data. He has published more than 70 papers in the area of survey sampling and missing data analysis. He is currently serving editorial boards of a number of scholarly journals, including the Annals of Applied Statistics, Statistica Sinica, Canadian Journal of Statistics, Survey Methodology, Annals of the Institute of Statistical Mathematics, and Journal of the Korean Statistical Society.

The workshop on machine learning with applications in the production of official statistics is divided into four parts. Examples of data science in the industry are first presented. The workshop continues with a presentation of the basic principles of machine learning, followed by a presentation of traditional methods of supervised and unsupervised learning. To conclude, applications of machine learning methods in the production of official statistics are discussed.

Jean-François Plante is an Associate Professor at HEC Montréal. He holds a PhD in Statistics from The University of British Columbia. His research programme aims to develop methodology for inference and predictions on distributed systems, a scalable architecture that is ubiquitous in the handling of big data. At HEC Montréal, Jean-François teaches data science and supervises the projects and theses of undergraduate and graduate students in that field. Jean-François is also a member of the Advisory Committee on Statistical Methods at Statistics Canada since October 2017.

The use of administrative data is ubiquitous in Statistical Agencies. This workshop gives the opportunity to participants to be exposed to methods that use these data in the context of official statistics, and more specifically for censuses. The workshop is split in four sessions. In the first session, participants will be presented an overview of methods that could be used to produce statistical registers of socio-economic characteristics and to assess their associated uncertainty based on administrative data. Then in the second session, the assumptions of the basic dual system estimator of population will be discussed. The methodological differences when one of the list is based on administrative data instead of a traditional census are highlighted. Erroneous enumeration in administrative registers will be the subject of the third session. Some of the most recent developments in modelling and treatment of erroneous and under enumeration in the available registers will be described, which can be applied at a detailed aggregation level. Finally, the workshop will end on the topic of census-like statistics of households and dwellings. Some relevant techniques for micro-integration of household and dwelling data, and the associated uncertainty assessment methods are presented.

Li-Chun Zhang is Professor of Social Statistics at University of Southampton, Senior Researcher at Statistics Norway, and Professor of Official Statistics at University of Oslo. He has worked and published on a number of subjects of interest to Official Statistics, such as sampling design and coordination, sample survey estimation, non-response, measurement errors, small area estimation, index number calculations, editing and imputation, register-based statistics, statistical matching, population size estimation.