R cheat sheet

Data manipulation Files & Objects

Vectors: Objects in memory: search() ls([n|”pth”|”regexp”]), find(“var”)
v<-scan() v<-1:n v<-c(1,3,4) v<-c(nm1=3,nm2=5,…) rm(ob) rm(list=ls()), library(), edit(obj),
v<-seq(lo,hi,step) v<-seq(length=n,from=lo,by=step) save(obj1,obj2,..,file=”/…”), ,save.image(file=””), load(fn)
v<-rep(expr,times=5) v<-rep(expr,each=5) File directories: getwd() setwd(“dir”) , unlink(“file”),
boolv=v>3 & v<5 strv=c(“vr”,”br”) list.files(path=””,pattern=””)
length(v) names(v) names(v)<-c(“p1”,”p2”..) Scripts: history(), save.history(file=””), edit(“file”),
v[1] v{2:4] v[v>3] v[boolv] v[-exclidxvec] v[c(“p1”,”p2)] sink(file=””,append=FALSE) for current coms,
factvec=factor(c(“f1”,”f2”,”f1”,…)) source(scrfn,echo=TRUE,print.eval=TRUE)
v=ordered(c(“fi”,”th”,),levels=c(“fi”,”se”)[,labels=c()]) Reading dataframes:
Matrices: df=read.table(file=”/..|URL”,header=FALSE,sep=””,
m<-matrix(v,nrow,ncol) m<-matrix(c(1,2,3,4,5,6),2,3) quote=”\”’”,row.names=strvec|colon|colname,na.strings=str
diag(vec) col(matrix) row(m) A[col(A)>=row(A)] vec, skip=n, strip.white=FALSE,blank.lines.skip=TRUE,
dim(m)<-c(row,col,..) dimnames(m)<-list(c(),c()…) comment.char=”#”) also read.csv(), write.table(obj,”file”)
m[vec,vec] m[,vec] m[vec,] matrix ops: %*% %/% %o% Graphics
eigen(sqrm) $values, $vectors svd(m) $u $v $d
plot(x,y[,type=”p[oint]|l[ines]|n[one]|b[oth]|s[tep]”,main=”title”,s
solve(m) solve(m,b)
ub=”subtitle”,xlab=””,ylab=””, logx|y|xy,
rowconcat(w/recycle): rbind(m1,m2) colconcat: cbind(m1,m2)
xlim=2vec,ylim=2vec)
aperm(m,c(newdim1,newdim2,…))
also: hist(x), bar(x), sunflowerplot(x,y) piechart(x), boxplot(x),
Lists/objects:
coplot(x,y|z), interaction.plot(f1,f2,y), qqplot(x), qqline(x)
li<-list(na1=obj,na2=obj,…) li<-c(li,na5=obj)
options: add=FALSE, axes=TRUE, type=”p”,”l”,”o”, xlim=v,
li$na1 li$na1[2] li[[2]] li[[2]][3]
ylim=, xlab=,ylab=,main=,sub=””, pch=0|..|25|”x”, pty=”s”|”m”
attributes(li) length(li)
lowlevel: points(x,y),lines(x,y),text(x,y,””), arrows(x0,y0,x1,y1),
objects: class(o) names(o) methods(fun) func(obj)= func.class(o)
abline(h=y,v=x,reg=mod,coef=v), legend(x,y,legend), rug(x)
Data frames: list of vectors with names, matrix index too
params: ?par, incl options, bg=”col”cex*=txtsz, col*=”color”,
df$fred df$fred[3] df[,”fred”] df[3:9,c(“fred”,”bob”)]
font*, las=txtor ,lty=1-6, lwd=width,mfcol(c(nr,nc)), ps=pts,
attach(df) detach(df) summary(df)
*= has .axis, .lab, .main, .sub also options above for par
df<-data.frame(vecs,factrs,..[row.names=,col.names=])
op<=par();par(…);par(op); par(mfrow=c(2,2) 2x2 mutliplot seq
crosstabbing: attach(df);table(fact1[,fact2],…) gives cts while
pairs(df) (matrix scatter) persp(x,y,z) contour(x,y,z) image(x,y,z)
tapply(var,fact|list(fac1,fac2),func) gives mean/std etc
3-D grid data
Programming windowmgt: x11(), pdf() dev.list() dev.cur() dev.set(n) dev.off(n)
Special values: NaN Inf –Inf TRUE FALSE Models
Operators: %% (modulo) %/% (int division)
formulas: var|log(v)~v1+v2-v3+v1:v2+v1*v3+log(v5)+I(v1*v2)
Assignment: <- -> = _ as in C returns val
-1 excludes constant, * gives interaction+indiv terms
Logical: !, & | vectorwise, &&, || scalar with partial evaluation
(var1+var2+var3)^2 poly(var1,2) or poly(var1,var2,3)
Typecasting:
To create models:
as(o,”numeric” |”character”|”logical”,”|vector” matrix”)
obj<-lm(formula,data=df,na.action=na.omit,subset=vec)
is(obj, as above), also as.numeric(), is.numeric()
obj<-glm(…,family=binomial|Gamma|Gaussian|poisson|
recycling: 2*c(1,2)+c(3,4,5,6)+1 = (6,9,8,11)
inverse.Gaussian)
Common functions: round trunc floor deiling, abs, sign, log,
library(nls);nls(Var~exprVar,data=,start=namvec,trace=T)
log10, sqrt, exp, sin, asin cosh log2 log(x,base) gamma lgamma
library(mgcv);obj<-gam(var~s(var1)+s(var2)+var3,…)
sum prod cumsum cumprod cummax cummin range
library(mda);obj<-mars(xmatrix,y,degree=n)
mean var std quantile(x,pvec) duplicated(c(1 2 1 4 2))=F F T F T
libarary(rpart);obj<-rpart(formula,…)
unique union intersect setdiff
library(nnet);obj<-nnet(…,size=n) obj<-lowess(x,y)
Functions:
To analyze: print, plot, summary, predict(obj,newdata), fitted,
name<-funct(unnam1,unarg2,named1=def1,…) {…. return(3); }
residuals, coefficients, plot(dep~indep)
call as: name(1,2) or name(na=2,1,2) or name(1,2,3)
for (i in vector) { } or stat; while (boolexpr) statmt Probability
repeat {} stmt; if (condition) statement else statement sample(vector,num) sample(1:n,num) density(x,bw=0.1)
ifelse(boolvec,trueval,falseval), d*(x,params,log=FALSE) gives prob(pdf) at x
apply(m,veckeepdims,function[,funcargs]) p*(x,params) gives CDF at x q*(x,params) gives inverse CDF
sweep(m,veckeepdims,var[,op=-]) does op using var r*(n,params) generates n random# fitdistr(x,”*”,knownparam=)
Strings: paste(v1,v2,sep=””,collapse=””) (collapse bet els) * may be: norm mean=,sd= lnorm meanlog=,sdlog=
substring(text,firt,last=1000000) beta shape1=,shape2= binom size=,prob=
grep(regexp,strvec) returns index of matches cauchy location=,scale= chisq df= exp rate=
regexpr(regexp,vec) returns –1 if false, posi if true f df1=,df2= gamma shape=,rate= geom. prob=
cat(“str1”,”str2”,”\n”) outputs string str=format(num,digits=n, hyper m=,n=,k= logis location=,scale=
justify=”left”|”right”, trim=TRUE, big.mark=”,”) nbinom size=,prob= pois lambda= t df=
unif min=,max= weibull shape=,scale= Wilcox m=,n=