### -----------------------------------------------------------------------
### Programmieren mit statistischer Software (SoSe 2017)
### -----------------------------------------------------------------------


## ----hsaur1--------------------------------------------------------------
# install.packages("HSAUR2")
library(HSAUR2)

## ----hsaur2--------------------------------------------------------------
data(package = "HSAUR2")

## ----hsaur3--------------------------------------------------------------
data("Forbes2000", package = "HSAUR2")
ls()

## ----hsaur44-------------------------------------------------------------
?Forbes2000
help("Forbes2000", package = "HSAUR2")

## ----hsaur5--------------------------------------------------------------
str(Forbes2000)

## ----hsaur6--------------------------------------------------------------
str(Forbes2000, vec.len = 1) # verkuerzter Ausschnitt

## ----hsaur7--------------------------------------------------------------
print(Forbes2000)   # alle Zeilen

## ----hsaur8a-------------------------------------------------------------
head(Forbes2000)    # die ersten 6 Zeilen

## ----hsaur8b-------------------------------------------------------------
tail(Forbes2000)    # die letzten 6 Zeilen

## ----hsaur9--------------------------------------------------------------
head(Forbes2000, n=2) #Ausgabe der ersten beiden Zeilen
tail(Forbes2000, n=10) #Ausgabe der ersten 10 Zeilen

## ----hsaur10-------------------------------------------------------------
nrow(Forbes2000)    # Anzahl an Zeilen
ncol(Forbes2000)    # Anzahl an Spalten
dim(Forbes2000)     # Anzahl an Zeilen und Spalten

# dim(Forbes2000)[1]
# dim(Forbes2000)[2]

## ----hsaur11-------------------------------------------------------------
names(Forbes2000)

## ----sum1----------------------------------------------------------------
summary(Forbes2000)

## ----sum2----------------------------------------------------------------
mean(Forbes2000$sales)
median(Forbes2000$sales)
range(Forbes2000$sales)
quantile(Forbes2000$sales)
quantile(Forbes2000$sales,probs=seq(0,1,by=0.05))

## ----sum3----------------------------------------------------------------
table(Forbes2000$category)
prop.table(table(Forbes2000$category))
nlevels(Forbes2000$category)
levels(Forbes2000$category)

## ----sub1----------------------------------------------------------------
Forbes2000[1:3, ]
Forbes2000[-(4:2000),]

## ----sub2----------------------------------------------------------------
Forbes2000[, 2] # nach Index
Forbes2000$name # nach Name
Forbes2000[, "name"] # nach Name
Forbes2000[,c("name", "country")]
Forbes2000[,-c(3:ncol(Forbes2000))] # nach Index
# Forbes2000[, -"name"] # funktioniert nicht
Forbes2000[,-which(names(Forbes2000) == "name")]
Forbes2000[,-which(names(Forbes2000) %in% c("name", "country"))]

## ----sub3----------------------------------------------------------------
vars <- c("name", "sales", "profits", "assets")
Forbes2000[1:3, vars]

## ----sub4a---------------------------------------------------------------
Forbes2000$assets > 1000

## ----sub4b---------------------------------------------------------------
table(Forbes2000$assets > 1000)

## ----sub5----------------------------------------------------------------
Forbes2000[Forbes2000$assets > 1000, "name"]
# mit which Zugriff auf den Index
Forbes2000[which(Forbes2000$assets > 1000), "name"]  

Forbes2000[Forbes2000$country == "Australia" & Forbes2000$rank < 100, ] # und-Operator

## ----sub5b---------------------------------------------------------------
Forbes2000[Forbes2000$country == "Australia" | Forbes2000$rank < 100, ] # oder-Operator

## ----sub6----------------------------------------------------------------
?subset
subset(x=Forbes2000, subset=Forbes2000$assets > 1000, select=vars)

## ----miss1---------------------------------------------------------------
summary(Forbes2000$profits)

## ----miss2---------------------------------------------------------------
mean(Forbes2000$profits)
?mean
mean(Forbes2000$profits, na.rm = TRUE)

cor(Forbes2000$profits, Forbes2000$sales)
?cor
cor(Forbes2000$profits, Forbes2000$sales, use="pairwise.complete.obs")

## ----miss3a--------------------------------------------------------------
which(is.na(Forbes2000$profits))

# which(is.na(Forbes2000))    # nicht so hilfreich
which(is.na(Forbes2000), arr.ind=TRUE)

## ----miss3b--------------------------------------------------------------
# Den Index der gueltigen Werte einer Variable erhaelt man mit
which(!is.na(Forbes2000$profits))

## ----miss4---------------------------------------------------------------
complete.cases(Forbes2000$profits)  # vollstaendige Eintr\"age
complete.cases(Forbes2000)          # vollstaendige Beobachtungen (Zeilen)

## ----miss5---------------------------------------------------------------
?na.omit
Forbes2000cc <- na.omit(Forbes2000)
dim(Forbes2000cc)
mean(Forbes2000cc$profits)

## ----manip1--------------------------------------------------------------
Forbes2000cc$costs <- Forbes2000cc$sales - Forbes2000cc$profits

## ----manip2--------------------------------------------------------------
Forbes2000cc$category <- NULL
dim(Forbes2000cc)

## ----manip3--------------------------------------------------------------
Forbes2000cc[, 4:7] <- scale(Forbes2000cc[, 4:7])

## ----manip4--------------------------------------------------------------
median(Forbes2000cc[, 4:7])

c(median(Forbes2000cc[, "sales"]),
  median(Forbes2000cc[, "profits"]))

## ----manip5--------------------------------------------------------------
?apply
apply(X=Forbes2000cc[, 4:7], MARGIN=2, FUN=median)

?sapply
sapply(X=Forbes2000cc[, 4:7], FUN=median)

## ----manip6--------------------------------------------------------------
Forbes2000cc[722, ]

## ----manip7--------------------------------------------------------------
Forbes2000cc[722, "sales"] <- 0.05
Forbes2000cc[722, ]

## ----save1---------------------------------------------------------------
?saveRDS
saveRDS(Forbes2000cc, file = "Forbes2000cc.rds")
rm(Forbes2000cc)

## ----save2---------------------------------------------------------------
Forbes2000cc <- readRDS("Forbes2000cc.rds")

## ----save3---------------------------------------------------------------
?write.table  # zum Speichern von Datensaetzen als Text-Datei
?save         # zum Speichern eines oder mehrerer Objekte
?save.image   # zum Speichern des gesamten Arbeitsverzeichnises

# setwd("...")
save.image(file="workspace.RData")