I'm using a modified version of the Cereals dataset from Chapter 28 the textbook as an example. The original dataset is available here: datasets from the text 

Cereals <- read.csv("~/Ch28_Cereals.csv")

Here's an idea of what the data looks like:

head(Cereals)
##                        name     brand IsKellogs calories sugars carbo protein
## 1                 100%_Bran   Nabisco        no       70      6   5.0       4
## 2         100%_Natural_Bran    Quaker        no      120      8   8.0       3
## 3                  All-Bran   Kellogs       yes       70      5   7.0       4
## 4 All-Bran_with_Extra_Fiber   Kellogs       yes       50      0   8.0       4
## 5            Almond_Delight   Ralston        no      110      8  14.0       2
## 6   Apple_Cinnamon_Cheerios Gen-Mills        no      110     10  10.5       2
##   fat sodium fiber potass shelf
## 1   1    130  10.0    280 Lower
## 2   5     15   2.0    135 Lower
## 3   1    260   9.0    320 Lower
## 4   0    140  14.0    330 Lower
## 5   2    200   1.0     -1 Lower
## 6   2    180   1.5     70 Upper

Numerical summaries

Mean, median, standard deviation, interquartile range (IQR)

Remove missing values with the na.rm option

mean(Cereals$sugars)
## [1] 6.986842
median(Cereals$sugars)
## [1] 7
sd(Cereals$sugars, na.rm=TRUE)
## [1] 4.428675
IQR(Cereals$sugars)
## [1] 8

5-number summary (Minimum, Q1, median, Q3, Max) 

summary(Cereals$sugars)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   0.000   3.000   7.000   6.987  11.000  15.000

Making subsets of datasets

Make a new dataset with only the brands that aren't Kellogs:

newdata <- subset(Cereals, brand !="Kellogs")

Make a new dataset with only Kellogs:

newdata <- subset(Cereals, brand =="Kellogs")

Make a new dataset with only the cereals with less than 100 calories:

newdata <- subset(Cereals, calories < 100)

Normal probability plot 

qqnorm(Cereals$sugars)
qqline(Cereals$sugars)

Correlation 

cor(Cereals$sugars, Cereals$carbo, use="pairwise.complete.obs") 
## [1] -0.4775783

Regression

lm stands for linear model.

lm(Cereals$sugars ~ Cereals$carbo)
## 
## Call:
## lm(formula = Cereals$sugars ~ Cereals$carbo)
## 
## Coefficients:
##   (Intercept)  Cereals$carbo  
##       14.9833        -0.5392
summary(lm(Cereals$sugars ~ Cereals$carbo))
## 
## Call:
## lm(formula = Cereals$sugars ~ Cereals$carbo)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -10.6693  -2.7060  -0.3554   3.5073   7.1839 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)    14.9833     1.7682   8.474 1.64e-12 ***
## Cereals$carbo  -0.5392     0.1153  -4.676 1.29e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3.917 on 74 degrees of freedom
## Multiple R-squared:  0.2281, Adjusted R-squared:  0.2176 
## F-statistic: 21.86 on 1 and 74 DF,  p-value: 1.286e-05
plot(jitter(Cereals$carbo), jitter(Cereals$sugars))
abline(lm(Cereals$sugars ~ Cereals$carbo))

Finding the residuals:

temp <- resid(lm(Cereals$sugars ~ Cereals$carbo))
plot(Cereals$carbo,temp)
abline(lm(temp ~ Cereals$carbo))


If you have a missing value, use the option na.action=na.exclude:

temp <- resid(lm(Cereals$sugars ~ Cereals$carbo, na.action=na.exclude))

Manipulating data

Remove an outlier 

Cereals$carbo[1] <- NA

Making an indicator variable (converting from categorical to quantitative)

In this new variable, "Kellogs" gets converted to 1 and other values get converted to 0.

Kellogscheck <- ifelse(Cereals$brand == "Kellogs", 1, 0)
table(Kellogscheck)
## Kellogscheck
##  0  1 
## 53 23

Making a factor variable (converting from quantitative to categorical) 

Cereals$protein <- as.factor(Cereals$protein)

Renaming or recoding categories in a categorical variable

install.packages("plyr") 
library(plyr)
## Warning: package 'plyr' was built under R version 3.6.3
levels(Cereals$brand) <- sub("^Ralston$", "Other", levels(Cereals$brand))
combinedCereals <- revalue(Cereals$brand, c("Ralston"="Other", "Quaker"="Other", "Post"="Other"))
barplot(table(combinedCereals))

Re-expression by square root or natural log 

sqrtfiber <- sqrt(Cereals$fiber)
logfiber <- log(Cereals$fiber)
boxplot(Cereals$fiber, sqrtfiber, logfiber)
## Warning in bplt(at[i], wid = width[i], stats = z$stats[, i], out = z$out[z$group
## == : Outlier (-Inf) in boxplot 3 is not drawn

Re-expressing and un-re-expressing invididual values:

log(10)
## [1] 2.302585
exp(2.302585)
## [1] 9.999999
sqrt(10)
## [1] 3.162278
(3.162278)^2
## [1] 10