With the advance of modern technology, more and more data are being recorded continuously during a time interval or intermittently at several discrete time points. These are both examples of functional data, which has become a commonly encountered type of data. Functional data analysis (FDA) encompasses the statistical methodology for such data. Broadly interpreted, FDA deals with the analysis and theory of data that are in the form of functions. This paper provides an overview of FDA, starting with simple statistical notions such as mean and covariance functions, then covering some core techniques, the most popular of which is functional principal component analysis (FPCA). FPCA is an important dimension reduction tool, and in sparse data situations it can be used to impute functional data that are sparsely observed. Other dimension reduction approaches are also discussed. In addition, we review another core technique, functional linear regression, as well as clustering and classification of functional data. Beyond linear and single-or multiple-index methods, we touch upon a few nonlinear approaches that are promising for certain applications. They include additive and other nonlinear functional regression models and models that feature time warping, manifold learning, and empirical differential equations. The paper concludes with a brief discussion of future directions.Annu. Rev. Stat. Appl. 2016.3:257-295. Downloaded from www.annualreviews.org Access provided by Academia Sinica -Life Science Library on 06/04/16. For personal use only. Section 2, and several approaches for dimension reduction in functional regression are discussed in Section 3. Clustering and classification of functional data are useful and important tools with wide-ranging applications in FDA. Methods include extensions of classical k-means and hierarchical clustering, Bayesian and model approaches to clustering, and classification via functional regression and functional discriminant analysis. These topics are explored in Section 4. The classical methods for FDA have been predominantly linear, such as functional principal components (FPCs) or the FLM. As more and more functional data are generated, it has emerged that many such data have inherent nonlinear features that make linear methods less effective. Sections 5 reviews some nonlinear approaches to FDA, including time warping, nonlinear manifold modeling, and nonlinear differential equations to model the underlying empirical dynamics.A well-known and well-studied nonlinear effect is time warping, where in addition to the common amplitude variation, one also considers time variation. This creates a basic nonidentifiability problem, and Section 5.1 provides a discussion of these foundational issues. A more general approach to modeling nonlinearity in functional data is to assume that the functional data lie on www.annualreviews.org • Review of Functional Data Analysis 259