########################################################################################
########################################################################################
#I. play with some data 
########################################################################################
########################################################################################

#	A. a number 
########################################################################################
X<-3
X
class(X)
str(X)

#	B. a word
########################################################################################
X<-"butIDontWannaCode"
X
class(X)
str(X)

#	C. a vector
########################################################################################
X<-c(1,2,3,4,5)
X
class(X)
str(X)

#	D. a mixed bag?
########################################################################################
X<-c(1,2,3,"I","still","dont","wanna","code")
X
class(X)
str(X)

#	E. see whats in your workspace 
########################################################################################
ls()

#	F. remove X, then look again
########################################################################################
rm(X)
ls()

########################################################################################
########################################################################################
#0. logistics--btw, '#' at start of line means the line is a comment
########################################################################################
########################################################################################
#A. where are you?
#   Note that, although it is a PC, the file-naming conventions are like unix/linux
#              e.g., "/" instead of "\" in full-path file names
#              also upper-vs-lower case matters! 
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
getwd()

#B. where would you like to be?   

#	1. what commands might be available to help us with directories? 
#      "apropos" can search for commands containing a string we provide
apropos("folder")  #darn, no commands with "folder" in them 
apropos("dir")     #yes!  "dir" was a good hint for things we can do with irectories

#	2. try for help on a command that looks good... precede the command with "?"
?dir.exists 

#	3. apply your chosen command and set the working directory 
print(dir.exists("D:/ibf")) #machine-specific to the PC used for demo

#		a. set the working directory 
if(dir.exists("D:/ibf")) {
	setwd("D:/ibf")  #again, machine-specific to the PC used for demo 
	getwd()
}
#		b. how to see the files?
apropos("files") #show commands with word "files"
?list.files      #get help on a command that looks good 

list.files()     #execute the command 
#			i. regular expression--ends with "r"
#			   nice summary at https://www.icewarp.com/support/online_help/203030104.htm
#list.files(pattern="*.r$")

#		c. check the new location
#      note that getwd() and print(getwd()) appear to have the same effect
#	   why?  any command--like getwd()--renders its result to the environment from which it was called....
#            in this case, the interactive terminal
print(getwd())


########################################################################################
########################################################################################
#I. get some data ...
########################################################################################
########################################################################################

#A. if youre NOT running online.... saras package installer/loader
########################################################################################
source("http://www.ceresanalytics.com/R_for_IBF_BootCamp/installLoadPkg.r")

#	1. package needed to read excel
installLoadPkg("psych")

#A. From the web, load this file: ________________
#	1. for ease of access, assign the filename to a character variable 
#	   this can be a web reference (as shown) or a filename using unix-like syntax, 
#      e.g. "C:/users/sara/myFile.csv"
myFilename<-"http://ceresanalytics.com/R_for_IBF_BootCamp/EDA_timeSeries.txt"

#	2. then read it in 
#      the read.delim functions first argument is file... 
#      you can either specify parameters by name, or place them in the order read

tryCatch({
	myData<-read.table(myFilename,sep="\t",header=TRUE)
}, error=function(e) {
	print(paste("YOU HAVE NO FILE NAMED",myFilename))
})

# Note the use of "<-" in assignments.  This is a good pracice!
# Why:  "=" is used for temporary assignments, such as passing parameters to functions
#       "==" is used to test equality of values 
# So youll always be able to search your code for assignments, passed parameters, and equivalence tests
# *if* you restrict usage of "<-","=", and "==" to each

#		a. find out about the cars dataframe you just read in
str(myData)

#		b. render it to your terminal
#		   when you just "blurt out" an object, it will be rendered to whatever environment
#		   from which its called 
myData 

#		c. what class is it
class(myData)

########################################################################################
########################################################################################
#II. Make a var for test/train, then create training and test datasets     
########################################################################################
########################################################################################

#	A. initialize as 1 (for "yes")
########################################################################################
myData$isTrain<-1
str(myData)

#	B. change to 1 for last 4 observations
########################################################################################
#		1. how many rows and columns?
#			a. rows 
nrow(myData)
#			b. columns 
ncol(myData)	
#			c. dimensions 
dim(myData)
#		2. use subsetting [] to isolate last 4 rows 
#			a. see it 
myData$isTrain
#			b. change it 
myData$isTrain[(nrow(myData)-3):nrow(myData)]<-0
#			c. see the data 
print(myData)

#	C. create training and test datasets 
########################################################################################
trainData<-myData[myData$isTrain==1,]
testData<-myData[myData$isTrain==0,]
#		1. check all dimensions
dim(trainData)
dim(testData)
dim(myData)

########################################################################################
########################################################################################
#II. Exploratory Data Analysis   
#    NOTE:  all data included, though a purist might look only at training data
########################################################################################
########################################################################################

#	A. what does a univariate line plot look like, as raw data?
########################################################################################
plot(
	x=myData$GMER,
	type="l"
)
#		1. for plotting, you can suppress the X axis and put in your own labels
#		   the catch is, just like a simple univariate plot, your X axis doesnt scale 
#		   see https://www.statmethods.net/advgraphs/axes.html
plot(
	x=myData$GMER,
	type="l",
	xaxt="n"
)
#		2. do it again, suppressing X-axis labels, then label the axes the way you want 
plot(
	x=myData$GMER,
	type="l",
	xaxt="n"
)
#			a. add the labels 
axis(1,
	at=1:length(myData$GMER), #which positions do you label 
	labels=myData$YearQuarterLabel, #what labels do you put there
	las=1 #use las=1 to put labels at right angles 
)

#	3. make a dep var into a time series
gmer_TS<-ts(myData$GMER, 
	frequency = 4,
	start = c(myData$Year[1], myData$Quarter[1])) # 1st obs year and quarter...

#	4. compare the structure of original data series with time series 													# assumes equal periodicity--which you can fill in with a SQL query
str(myData$GMER)
str(gmer_TS)

#	5. plot it ... dont tell it "l" this time 
plot(x=gmer_TS)
#		i. whats special about ts plot?  synatax of plot command is plot.class, e.g. plot.ts
class(gmer_TS)

#	6. make 3 series in the data frame each into time series  
myData<-myData
myData$FURN<-ts(myData$FURN, 
	frequency = 4,
	start = c(myData$Year[1], myData$Quarter[1])) #as above 
myData$FURN_highgrowth<-ts(myData$FURN_highgrowth, 
	frequency = 4,
	start = c(myData$Year[1], myData$Quarter[1])) #as above 
myData$FURN_goesToZero<-ts(myData$FURN_goesToZero, 
	frequency = 4,
	start = c(myData$Year[1], myData$Quarter[1])) #as above 

#	7. now look at 3 at a time... 

#		#i. min and max for uniform presentation 
myYMin<-min(c(myData$FURN,myData$FURN_highgrowth,myData$FURN_goesToZero))
myYMax<-max(c(myData$FURN,myData$FURN_highgrowth,myData$FURN_goesToZero))

#		#ii. "par" opens your plot window, then "mfrow" is how many 
par(mfrow=c(1,3))  
	plot(myData$FURN,ylim=c(myYMin,myYMax),main="Furniture sales, base case")
	plot(myData$FURN_highgrowth, ylim=c(myYMin,myYMax), main="Furniture sales, high case")
	plot(myData$FURN_goesToZero, ylim=c(myYMin,myYMax), main="Furniture sales, low case")

#	B. what do potential predictors look like?
########################################################################################

#		1. in these data, potential predictors are WASA, EMPL, QTR_4 and hypoInterestRate
predictorNames<-c("WASA","EMPL","QTR_4","hypoInterestRate")
#			a. does this vector work for subsetting?
head(myData[,predictorNames])
#			b. can you concatenate this vector with the predictor subsest?
depVarName<-"FURN"
head(myData[,c(depVarName,predictorNames)])

#		2. look at your predictors, each on their own
describe(myData[,predictorNames])
#			look at each predictors standard deviation relative to its mean
#			see how hypoInterestRate is virtually constant?

#		3. look at your predictors vs. the dependent variable
#		   youll again concatenate the dependent variable (FURN) with the predictor names (WASA, EMPL, QTR_4 and hypoInterestRate)
pairs(myData[c(depVarName,predictorNames)],pch=19,lower.panel=NULL)