Skip to content

Toggle service links

You are here

  1. Home
  2. Research
  3. Research groups
  4. Statistics
  5. Multivariate Statistics
  6. Multivariate methods with more variables than observations

Multivariate methods with more variables than observations

Chart showing mechanical aptitude score against verbal aptitude score for three groupings

This project aims to extend the classical multivariate techniques to situations where one needs to analyse data with more variables than observations. Such data are common in a number of important modern applications (for example meteorological data on climate, gene sequencing, and financial data analysis) but the standard multivariate techniques are either inefficient or simply cannot be applied.

Dr. Nickolay Trendafilov is currently funded by the Leverhulme Trust on a project entitled "Sparse factor analysis with application to large data sets"..