#By Stefan Zeugner, 2007-02-12
#R commands used in TA sesssion 1 on R
# QUICK INTRO INTO R FOR PROGRAMMING NEWBIES

#____ Contents: ______________________

# o BASIC INTRODUCTION
# o GENERATE VECTORS , MATRICES
# o WORKING WITH LISTS
# o CLEANING UP OBJECTS
# o RANDOM NUMBERS AND MATRIX ALGEBRA
# o CLASSES (variable types)
# o READING FILES AND DATA.FRAME
# o ESTIMATING
# o PREPARE DATAFRAME FOR SAVING
# o SAVE TO FILE
#______________________________________



#------- BASIC INTROCDUTION -----------

# define variables (here, money equation example):
 M<-50000  	# money supply
 100->P	# price level
 Y=1000	# real GDP
 v=P*Y/M 	# calculate velocity
 # (scalar operators: * / - + %% %/% ^ )

 print(v) # display the value of v

 foo="Milton" # define string / character variable

 ls() #list the workspace: all the variables (and other objects) defined so far
 ls # ls is a function; executing it requires brackets (); without the brackets, the object / function code is displayed, but not executed
 
 rm(M) #removing variables
 ls() # now M is not in the workspace anymore

 help(ls) #show help on ls function
 ?ls # equivalent command
 help.search("ls")

#create basic vector
 test=1
 test[3]=2 # add element
 test #display

#MODE TYPES: numeric(double and integer), character, logical, (raw, complex), list, function
 mode(test)
 mode(foo)

#logical operators == < >= !=, & | !
 3==3 #returns TRUE (3 equals 3)
 3!=3 #returns FALSE (it is false that "3 is not equal to 3")
 3=3 #returns error; because single "=" is used for assignment (equivalent to "<-")

# playing with modes
 bool = 4>3
 bool2 = FALSE
 mode(bool)
 bool[2]="dsf"
 mode(bool)


##------- GENERATE VECTORS , MATRICES -----------

#____operators c(), rbind, cbind ______
# c() combines stuff in one dimension (vector)
 c(4,3)
 c(1:3)
 c(10:-1)
 cumsum(c(1:10)) #cumsum does the cumulative sum of a vector

# cbind/rbind combine stuff along the column/row dimension in 2 dimensions
 cbind(1:2,3:4)
 cbind(c(1,2),c(3,4))
 mymatrix=rbind(c(1,2),c(3,4))
 print(mymatrix)

#___ other stuff _____
# construct vectors: 
 seq(from=10,to=2,by=-2) #equivalent to: seq(10,2,2)
 rep(1,10)
 rep("test",10,sep="_") 

#constuct matrix
 matrix(1,3,3)
 matrix(1:9,3)
 testmat=matrix(1:12,3,4)

#accessing elements
 mymatrix[2,1]
 mymatrix[,1]

#reshape
 matrix(testmat,2,6) 

#some operations on matrices
 mymat=matrix(1:9,3,3)
 length(mymat) # no. of elements
 dim(mymat)
 max(mymat)
 sum(mymat)
 det(mymat)
 eigen(mymat)

#array(): alternative to matrix()
 matrix(1:4,3,3)
 array(1:4,dim=c(3,3))
 mymat[,2]<-c(1,2,3)
 pmax(10:1,1:10)


#comparing vectors
 a=1:10
 b=rep(5,10)
 h<-a==b
 print(h)


#------- WORKING WITH LISTS -----------

#create a list (object)
 mylist=list(bool,bool2)
 # a list is an object that can hold all kinds of other objects: vectors, other lists, functions, etc.

 mode(mylist) #assess modes
 mode(mylist[[1]])
 mode(mylist[[2]])
 
 ls(pat="bool") 3 search for objects with name 'bool'

#manipulating lists 
 newlist=c(mylist,P)
 newlist
 newlist=list(mylist,P)
 newlist
 c(newlist,recursive=TRUE)
 mode(unlist(newlist))

#accessing elements
 otherlist=list(hihi=2,huhu=2:3,haha=matrix(1,3,3))
 names(otherlist)
 otherlist$haha 
 # the same as:
 otherlist[[3]]

# list and names!
 names(newlist)=c("a","b")
 newlist$a
 newlist$b

#------- CLEANING UP -----------
 ls(pat="mat")
 ls(pat="list\|mat") #list everything that is called with 'list' or 'mat'
 rm(list=ls(pat="mat\|list")) #delete everything that is called with 'list' or 'mat'
 rm(list=ls()) #remove everything


#------- RANDOM NUMBERS AND MATRIX ALGEBRA -----------
 #simulate data and do OLS on it:
 #matrix operators: t:transpose, solve:inverse, %*%: matrix mult 

 N=100 #no.obs.

#draw X matrix (normally distributed, 3 columns):
 matrix(rnorm(N*3),N,3) 

# draw error terms:
 e=rnorm(N,0,2) 

#make dep. variable:
 y=10*X[,1]-3*X[,2]+5*X[,3]-10+e 

# add a constant to the design matrix:
 design=cbind(rep(1,N),X) 

# calc coefs. b=(X'X)^(-1) X'y
 b=solve(t(design) %*% design) %*% t(design) %*% y

# do ehat=[I - X(X'X)^(-1)X'] y
 ehat=(diag(N)-design %*% solve(t(design) %*% design) %*% t(design)) %*% y
 print(b)





#------- CLASSES (variable types) -----------
#class "numeric", "logical", "character" or "list", but also "matrix", "array", "factor" and "data.frame" a
# "mode" are general types, "class" is a finer specifiation
 mymat=matrix(1:9,3,3)
 class(mymat)
 mode(mymat)


#------- READING FILES AND DATA.FRAME -----------

#set working directory to the directory where you have your files stored 
 setwd('C:/Dokumente und Einstellungen/zeugner/Eigene Dateien/IHS') 
#IMPORTANT: in your path, use forward slashes "/" instead of backward slashes "\" - also under Windows!

#read data from file
 usdata=read.csv("u_cpi_usa.csv") # consider also read.table("u_cpi_usa.csv",sep=",")

 class(usdata) #data.frame: a restricted list


#accessing elements of data.frame
 names(usdata)
 usdata$usa_cpi 

#use 'attach' for easy handling
 attach(usdata) #tell the workspace to directly access usdata list elements
 usa_cpi

#if this does not work use the following
 detach(usdata) #delink usdata list elements from workspace
 attach(usdata) #link usdata list elements from workspace


#summarize info
 summary(usa_infl)
 summary(usdata)


#calculate inflation
 infl=c(NA,diff(log(usdata$usa_cpi)))

#add inflation to data frame
 usdata=cbind(usdata,infl)
 names(usdata)[4]="usa_infl"

# assign the rownames
 dimnames(usdata)[[1]]<-date 

# drop, 1st and 3rd column of usdata
 usdata<-usdata[,-c(1,3)] 

#you have to do the attaching again to update information
 detach(usdata)
 attach(usdata)
 usa_cpi
 usa_infl

#------- ESTIMATING -----------
# lm() does OLS:
 lm(usa_unemp ~ usa_infl)
 #glm(usa_unemp ~ usa_infl,family="quasipoisson")

#assign results to object "model"
 model=lm(usa_unemp ~ usa_infl)
 class(model)
 names(model)
 summary(model)

# access attributes of lm object
 model$residuals 

#------- PREPARE DATAFRAME FOR SAVING -----------

hats=data.frame(ehat=model$residuals,yhat=model$fitted.values,row.names=row.names(usdata)[2:length(row.names(usdata))])

#we have to use row.names(usdata)[2:length(row.names(usdata))] as there is in NA in the first entry of usa_infl

#------- SAVE TO FILE -----------

 write.table(hats,"hatsout.txt")

#write to file with comma separator:
 write.table(hats,"hatsout.csv", sep=",")
 
#exit from R:
 q() # or q(save="no")

#by Stefan Zeugner, 2007