Postdoc at Rennes on multilevel missing data imputation

Julie Josse sends along this job announcement:

A post-doctoral position is available in the applied mathematics department of Agrocampus Rennes. The postdoc will be funded by the Henri Lebesgue Center (see http://www.lebesgue.fr/) if the application is selected. Applicants are expected to send their application before 31 March 2014.

The research focus is on development of new methods to deal with missing values and their implementation in the free R software to make them available. We study new multiple imputation methods based on principal component methods.

Different aspects are expected to be covered: dealing with missing values in multi-blocks, multi-groups data (groups of individuals and variables); regularization in this framework using a Bayesian approach, dealing with different types of data (continuous, categoricals, etc.). Fields of application are wide and include biological data as well as socio-economic data.

Key words: missing values, matrix completion, PCA, Bayesian PCA, dimensionality reduction methods, exploratory multivariate data analysis methods, correspondence analysis, multi-groups, multi-blocks methods, singular values thresholding, low rank approximation.

Profile
– PhD in Statistics
– Ability for programming in R, eventually in C/C++
– Good level of English

Information
Duration: 12 months
Salary: 2 225 € after taxes /month
Advantages: Help for housing

For more information please contact:
Julie Josse, Assistant Professor, [email protected]
François Husson, Professor, [email protected]
Applicants are requested to send CV and one letter of recommendation.