## R-Einfuehrung NDK
## --------------------------------------------------------------
## Daten einlesen
d.sport <- read.table("sport.dat",header=TRUE)
d.sport
d.sport[,"kugel"]
d.sport[5,]
## Grafik -- Funktionsaufruf
hist(d.sport[,"kugel"])
hist(d.sport[,"kugel"], nclass=10)
hist(d.sport[,"kugel"],,,,,,,,,,,,,,,,10)
objects()
ls()
# q()

## Grafik
plot(d.sport[,"kugel"], d.sport[,"speer"])
pairs(d.sport)
# dev.print()

## Hilfe
help(hist)
?hist
example(hist)
## Fehlermeldungen
## --------------------------------------------------------------
## Vektoren
t.v <- d.sport[,"kugel"]
t.a <- c(3.1, 5, -0.7, 0.9, 1.7)
seq(0,3,by=0.5)
1:9
rep(0.7,5)
rep(c(1,3,5),length=8)

## Alphanumerische Vektoren
t.b <- c("Andi" "Bettina", "Christian")
paste("ABC","XYZ",17)
paste("ABC","IJK","XYZ",sep=":")
paste(c("a","b","c"),1:3)
paste(letters, collapse="; ")
# paste hat sie beliebig viele Argumente!

## Logische Vektoren
(1:5)>=3
t.i <- (t.v>2)&(t.v<5)
## --------------------------------------------------------------
## Arithmetik
2+5
(2:5)^c(2,3,1,0)
(2:5)^2
(1:5)-(0:1)

## --------------------------------------------------------------
## Elemente auswählen
t.v[c(1,3,5)]
d.sport[c(1,3,5),1:3]
t.a[c(TRUE,FALSE,TRUE,TRUE,FALSE,FALSE)]
d.sport[c("OBRIEN","DVORAK"),c("kugel","speer","punkte")]
d.sport$kugel
        
## --------------------------------------------------------------
## Verteilungen und Zufallszahlen
dbinom(0:5, size=5, prob=0.7)
pbinom(0:5, size=5, prob=0.7)
qbinom(seq(0.1,0.9,0.1),size=5,prob=0.7)
rbinom(20, size=5, prob=0.7)
rpois(20, lambda=3.5)
runif(4)
runif(2)
set.seed(27)
runif(4)
set.seed(27)
runif(2)

plot(0:15,dpois(0:15,lambda=3.5), type="h", lwd=3)
t.x  <- seq(0,10,length=100)[-1]
plot(t.x,dnorm(t.x, 5, 2), type="l", xlab="x", 
  ylab="Dichte", main="Normalverteilung")
plot(t.x,dlnorm(t.x, log(3), log(1.7)), type="l", 
  xlab="x", ylab="Dichte", main="Log-Normalvert.")

## --------------------------------------------------------------
## Einfache Statistik-Funktionen
binom.test(3,20, p=0.4)
t.x <- 5
1-ppois(t.x-1,lambda=2.7)
t.hh <- d.sport[,"hoch"]>200
t.kugel <- d.sport[,"kugel"]
wilcox.test(t.kugel[t.hh],t.kugel[!t.hh])

## Grafische Darstellungen für zwei Stichproben.
boxplot(t.kugel[t.hh],t.kugel[!t.hh],notch=TRUE)
t.br <- 13:17
t.h1 <- hist(t.kugel[t.hh],breaks=t.br, plot=FALSE)
t.h2 <- hist(t.kugel[!t.hh],breaks=t.br, plot=FALSE)
barplot(rbind(t.h1$density,t.h2$density), beside=TRUE, names.arg=t.h1$mid)

## --------------------------------------------------------------
## Funktionen
  f.maxi <- function(data, na.remove=TRUE){
  l.max <- max(data, na.rm=na.remove)
  l.i <- match(l.max, data)
  c(max=l.max, i=l.i)
}

f.maxi(c(3,4,78,2))

apply(d.sport, 2, f.maxi)

## if

f.log <- function(x){
  if (any(x<=0)) {
    stop("Zahlen sind unbrauchbar.")
  } else {
    l.x <- log(x)
    return(l.x)
  }
}

x <- 1
for (i in 1:10){
  x <- x*(x+1)
  if (x>1000000) break
}

## Fehlermeldungen

traceback()}

f.maxi <- function(data, na.remove=TRUE){
  on.exit(browser())
  l.max <- max(data, na.rm=na.remove)
  l.i <- match(l.max, data)
  on.exit()
  c(max=l.max, i=l.i)
}

debug(f.maxi)

## Kapitel Objekte

## Listen

t.l <- hist(t.kugel,plot=FALSE)
t.l
## Komponenten einer Liste auswählen
t.l[2:3]
t.l[c("breaks","intensities")]
t.l["counts"]
t.l[[2]]
t.l[["counts"]]
t.l$counts


## data.frame
## Matrizen

t.m <- matrix(1:15, nrow=3, ncol=5)
t.m[2,1:3]

cbind(4:6,13:15)

## Eigenschaften von Objekten
mode(t.kugel)
is.matrix(d.sport)
is.data.frame(d.sport)
attributes(d.sport)
str(t.l)

t.sport.mat <- as.matrix(d.sport)
unlist(list(a=1:2, b=5:7))

## Objektorientiertes Programmieren
  
t.tt <- t.test(t.kugel[t.hh],t.kugel[!t.hh])
class(t.tt)
names(t.tt)

str(tt)
print(t.tt)
plot(t.tt)
summary(t.tt)

## libraries

library(MASS)
help.search("Wilcoxon")
## http://www.r-project.org/

## Weitere Informationene über ein Package
library(help=MASS)