ABSTRACT
This chapter focuses on the process of obtaining the model parameters that "best represent" the observed data. It discusses problems involving regression in single variable. The chapter explores few examples to highlight the procedure of parameter estimation. It also discusses the problem of fitting a straight line to observed data and statistical properties related to goodness of fit. The chapter describes nonlinear regression, including conversion of a nonlinear regression problem to linear regression. Linear least squares described implicitly assumes that the measurement errors draw from a normal distribution. An approach to qualitatively determine the goodness of fit is to plot the original data and model fit on the same figure. MATLAB has several solvers for linear and nonlinear regression. However, most solvers require a toolbox to be installed. The Statistics and Machine Learning Toolbox in MATLAB gives a broad spectrum of tools for data analysis, regression, and hypothesis testing.
