تحليل داده هاي تابعي: تعميمي از تحليل داده هاي چند متغيره

Functional Data Analysis: A Generalization of Multivariate Data Analysis

Many phenomena such as temperature and precipitation are essentially continuous functions of time. Although, it is possible to use the classical multivariate methods for analyzing the data, it may not be entirely correct . Because the data are functions, they must be considered in function spaces that are of infinite dimension, instead of consideration in vector spaces with finite dimension. In this way, some theoretical justification for Functional Data Analysis (FDA) is needed. This action leads to new statistical approaches that called FDA. In this article, we give definitions and concepts related to FDA with necessity of using them, the descriptive statistics for functional data, and comparison of functional data with both multivariate and longitudinal data. Finally, we discuss functional principal components analysis and functional linear regression.