Exploring the work space

• ls() # Lists the objects in the work space
  
• library() # Shows the installed packages
  
• dir() # show files in the working directory

Missing data

• rowSums(is.na(mydata)) # Number of missing per row
  
• colSums(is.na(mydata)) # Number of missing per column/variable
  
• length #Num. of variables/elements in an object
  
• mydata[mydata$age=="& ","age"]<- NA # Convert to missing data
  
• mydata1 <- na.omit(mydata) # Creating a new data set without missing data

 Replacing a value

• mydata1 <- na.omit(mydata)
  
• mydata1[mydata1$SAT==1787,"SAT"] <- 1800
  
• mydata1[mydata1$Country=="Bulgaria","Country"]<- "US"

Renaming variables

• library(reshape)
  
• data <- rename(data, c(Relationship.Status="Status"))
  
• data <- rename(data,c(Average.score..grade.="Score"))
  

Value labels

• mydata$sex <- factor(mydata$sex, levels = c(1,2), labels= c("male", "female")) # Use factor() for nominal data
  
• mydata$var2 <- ordered(mydata$var2, levels =c(1,2,3,4), labels = c("Strongly agree", "Somewhat agree", "Somewhat disagree","Strongly disagree")) # Use ordered() for ordinal data

Reordering labels

• levels(mydata$Major) # Syntax for reorder(categorical variable, numeric variable, desired statistic)
  
• mydata$Major = with(mydata, reorder(Major,Read,mean)) #Order goes from low to high

Creating ids/sequence of numbers

• mydata$id <- seq(dim(mydata)[1]) # Creating a variable with a sequence of numbers from 1 to n (where ‘n’ is the total number of observations)
  
• mydata$total <- dim(mydata)[1] # Creating a variable with the total number of observations

Recoding variables

• library(car)
  
• data$qty.rec <- recode(data$quantity, "1:9='1to9'; 10:19='10to19'; 20:29='20to29'")
  
• data$ qty.rec <- as.factor(data$ qty.rec)

Sort

• data.sorted <- data[order(data$Gender),] 
  
• data.sorted1 <- data[order(data$Gender, -data$SAT),]
  

Deleting variables

• data$qty.rec <- NULL 
  

Sub-setting variables

• select1 <- data[,1:14] # Selecting the first 14 variables
• data2 <- data2[1:30,] # Selecting the first 30 observations
• mydata2 <- mydata2[1:30,1:14] # Selecting the first 30 observations and first 14 variables
• sat <- data[c("Last", "First", "SAT")] # Selecting specific variables
  
• select2 <- data[c(1:3, 12:14)] # Type names(select) to verify
  
• select3 <- data[c(-(1:3), -(12:14))] # Excluding variables
• data3a <- data[which(data$Gender=='Female' & data$SAT > 1800), ] # Selecting the specific observations
  
• Data1 <- subset(data, Age >= 20 & Age <= 30)
  
• Data2 <- subset(data, Age >= 20 & Age <= 30, select=c(ID, First, Last, Age))
  
• Data3 <- subset(data, Gender=="Female" & Status=="Graduate" & Age >= 30)
  

Categorical data: Frequencies/Cross-tabs

• table(mydata$Gender) # One-way tables
  
• readgender <- table(mydata$Read,mydata$Gender) # Two-way tables
  
• addmargins(readgender) # Adding row/col margins
• prop.table(readgender,1) # Row proportions
• prop.table(readgender,2) # Column proportions
• prop.table(readgender) # Tot proportions
• round(prop.table(readgender,1), 2) # Round col prop to 2 digits 
• chisq.test(readgender) # Do chisq test Ho: no relathionship 
• fisher.test(readgender) # Do fisher'exact test Ho: no relationship
• table3 <- xtabs(~Read+Major+Gender, data=mydata) # 3-way crosstabs

Descriptive Statistics

• library(fBasics) #calling used library
  
• basicStats(data$sqft.living) #descriptive for one variable

• library(psych) #calling used library
• describe.by(data$sleep, data$group) #describing sleep by group

Graphs

• barplot(margin.table(readgender,1))  # Plotting frequencies
  
• boxplot(data$bedrooms, main="bedrooms")  #boxplot
• boxplot(birth~Feed,data=book1, main="boxplot", xlab="Feed", ylab="birthweight") #stacked boxplot
  
• hist(mydata$SAT, breaks="FD", col="green") # Histograms overlaid
  
• plot(x, y, xlab="x", ylab="y", main="Scatter Plot")  #scatterplot
  
• pairs(~ x1+ x2+ x3+ x4, span=0.7, id.n=0, data=mydata) # Scatterplots (multiple)
  
• pairs(mydata) # Pariwise plots
  
• plot(vocabulary ~ education, data=Vocab) # Scatterplots (for categorical data)

Correlation and Regression

• cor.test(~price+sqft.living, data=mydata) # Correlation

• mydata$new.variable <- ifelse(mydata$original.variable=='Yes', 1, 0) #create dummy variable

• fit <- lm(price~lnsqft.living, data=mydata) #Regression
  
• summary(fit)  #view regression results
  
• resid(fit)  #calculate residuals

• plot(density(resid(fit)))  #plot PP plot of residuals
  
• qqnorm(resid(fit))  #plot QQ plot of residuals
  
• plot(savings, resid(fit), xlab="personal savings", ylab="residual", main="Residual Plot") #plots residual plot

• newdata = data.frame(x=21) #prediction for a particular value of independent variable 
• predict(fit, newdata, interval = "prediction") #prediction for a particular value of independent variable 
• predict(fit, newdata = new.speeds, interval = "prediction") #prediction interval for a set of values of independent variable
• predict(fit, newdata = new.speeds, interval = "confidence") #confidence interval for a set of values of independent variable

Hypothesis Testing