Preface to First Edition - lib

16 AN INTRODUCTION TO Rto light when writing our own functions for our own analysis tasks. AlthoughR is a full-featured programming language, writing small helper functions forour daily work is not too complicated. We’ll study two example cases.At first, we want to add a robust measure of variability to the locationmeasures computed in the previous subsection. In addition to the medianprofit, computed viaR> median(Forbes2000$profits, na.rm = TRUE)[1] 0.2we want to compute the inter-quartile range, i.e., the difference betweenthe 3rd and 1st quartile. Although a quick search in the manual pages (viahelp("interquartile")) brings function IQR to our attention, we will approachthis task without making use of this tool, but using function quantilefor computing sample quantiles only.A function in R is nothing but an object, and all objects are created equal.Thus, we ‘just’ have to assign a function object to a variable. A functionobject consists of an argument list, defining arguments and possibly defaultvalues, and a body defining the computations. The body starts and ends withbraces. Of course, the body is assumed to be valid R code. In most cases weexpect a function to return an object, therefore, the body will contain one ormore return statements the arguments of which define the return values.Returning to our example, we’ll name our function iqr. The iqr functionshould operate on numeric vectors, therefore it should have an argument x.This numeric vector will be passed on to the quantile function for computingthe sample quartiles. The required difference between the 3 rd and 1 st quartilecan then be computed using diff. The definition of our function reads asfollowsR> iqr IQR(xdata)[1] 1.495980However, when the numeric vector contains missing values, our function failsas the following example shows:R> xdata[1] iqr(xdata)© 2010 by Taylor and Francis Group, LLC