This PhD thesis presents methods for the analysis of life course data in demography and social sciences. Their focus is the analysis of categorical sequences representing life trajectories, that is, series of statuses or roles that an individual enacts over time in domains such as education, family or work. The presented methods aim to discover, describe and explain life course patterns. A first contribution is anchored in the main existing framework for sequence analysis, based on the computation of pairwise dissimilarities between sequences. It addresses the issue of summarizing sets of life trajectories with typical patterns. The second contribution proposes a general framework for analyzing life trajectories that is based on a particular class of markovian models called variable length Markov chains.