Obj <- or =number/characters
Vector <- c(number, “value”)
Length(V) :
Min/max/range/mean/sum(V)
Setwd(“location”)
Dataframe <- read.csv(“file name”)
Summary(df): simple stats of each variable
#(min quartile median mean max)/N(length)
Head(df):first six observations
Tail(df): last six observations
Head(df$variable):first 6 obs. for variable
Same for mean/table – frequencies
Save file:
write.csv(resume, file="resume1.csv")
save(resume, file="resume1.RData")
!not &and |or!=not equal>=大于等于<=小于等于
==logical operator (e.g.5==6 false)
!=not true (e.g. 5!=6 True)
newV <- V2>V1 drop obs. V2>V1
df<- c(1,2,3,4)
df(3) = 3/df(-3) = 1 2 4
[] indexing:
df[c(1,3)] = 1 3
df[c(TRUE, FALSE, TRUE, FALSE)]
= 1, 3 (only 1st & 3rd is true)
school <- c("UCSD", "UCB", "UCLA", "UCR")
school=="UCSD"
[1] TRUE FALSE FALSE FALSE
Subset graduationdate vector
graduationdates[school=="UCSD"]
[1] 2010
Conditional vector
Vec <- df$variable == “condition”
Frequency of conditional vector
Sum(df$variable==“condition”)
= # of obs.
Mean of V1 with the condition of V2
Mean(df$variable[df$variable2])
Subsetting data frames
students <- data.frame(school=c("UCSD",
"UCB", "UCSD"),
graduationdate=c(2010, 2019, 2015))=
students
school graduationdate
1 UCSD 2010
2 UCB 2019
3 UCSD 2015
Specify data frame, row, column:
Students[3,1] / [1,] / [,2]
[1] “UCSD” / 1 UCSD 2010/2010 2019 2015
Extract row 1&3 w/o 2
students[c(1,3),] (or use students[-2,])
school graduationdate
1 UCSD 2010
3 UCSD 2015
Extract where ==“UCSD” is true
students school=="UCSD"
[1] TRUE FALSE TRUE
⬆️(return logical factor)⬇️ filter
students[students$school=="UCSD",]
school graduationdate
1 UCSD 2010
3 UCSD 2015
Tidyverse: subset(df,logical statement)
subset(students,students$school=="UCSD")
school graduationdate
1 UCSD 2010
3 UCSD 2015
Create dataframe for subset
resume_blacknames <-resume[resume$race=="black",]
resume_whitenames <-
resume[resume$race=="white",]
mean of subset df
mean(resume_blacknames$variable)
[1] 0.06447639
CONDITIONAL STATEMENTS
if (logical statement) {
code to be executed if logical statement is TRUE
}
EG) door <- "locked"
if (door=="locked") {
print("sorry, you need a key to enter")
}
[1] "sorry, you need a key to enter"
EG2) door <- "unlocked"
if (door=="locked") {
print("sorry, you need a key to enter")
} NOT EXECUTED AS DOOR LOCKED NOT TRUE
ELSE EG) door <- "locked"
if (door=="locked"){
print("Sorry, you need a key to enter")
} else { {and else same line must
print("Please Come in!")
}
[1] "Sorry, you need a key to enter"
Executes else if door<-“unlocked”
[1]”please come in!”
Loops:
for (some set of things) {
do some stuff
}
EG) i <- 2
print(2*i)
[1] 4
EG2) for (i in c(3,10,99)){
print(2*i)
}
[1] 6 [1] 20 [1] 198
EG3) homework <- c("math", "reading", "writing")
for (i in homework) {
cat("Do", i, "\n")
}
Do math
Do reading
Do writing
*cat(): concatenates & prints
*\n: display in next line
Same process by looping over words in hw:
for (i in 1:length(homework)) {
cat("Do", homework[i], "\n")
}
Dimensions: Dim(df)
Subset individuals (age>=25 & age<=34 & mother2==1):
mothers2534 <- subset(df, mother2==1 & age>=25 &
age<=34)
install tidyverse
install.packages("tidyverse")
library(tidyverse)
convert data frame to tibble:
rr <- as_tibble(rr)
retrieves rows of data that meet certain condition:
filter(dataframe, some logical statement)
EG) mothers2534 <- filter(rr, age<=34 & age>=25 &
mother2==1)
Or EG) filter(mothers2534, dataset%in%2003:2008)
Pipe eg) mothers2534 <- rr %>% filter(age<=34 &
age>=25 & mother2==1)
Select remain tibble wanted:
Eg) mothers2534 <- select(mothers2534, dataset,
mother2, age, childtot)
Sort data based on values of variable:
mothers2534 <- arrange(mothers2534,age)
want descending order:
mothers2534 <- arrange(mothers2534,desc(age))
head(mothers2534$age)
[1] 34 34 34 34 34 34
Pipe operator:
F(x) = x%>% f()
For multiple functions:
h(g(f(x)))
x %>%
f %>%
g %>%
1st filter 2nd selected variables needed:
mothers2534 <- rr %>%
filter(age<=34 & age>=25 & mother2==1) %>%
select(dataset, mother2, age, childtot)
*if want descending order add:
%>%
arrange(desc(age))
new variable
df$newvarname <- expression
eg) rr$childcollegeprep <- rr$childeduc + rr$childtravel
or: mutate(dataframe, newvarname = expression)
eg) rr <- mutate(rr,
childcollegeprep=childtravel+childeduc)
can create mutiple new var at same time:
rr <- mutate(rr, childcollegeprep=childtravel+childeduc,
childnotcollegeprep=childtot-childcollegeprep)
create new var, drop all prior var:
collegeprepdat <- transmute(rr,
childcollegeprep = childeduc + childtravel,
childnotcollegeprep = childtot - collegeprep)
Error in `transmute()`:
ℹ In argument: `childnotcollegeprep = childtot -
collegeprep`.
Caused by error:
! object 'collegeprep' not found
Summarize() – generate summary stats:
summarize(rr, meanchildtot = mean(childtot, na.rm=T))
# A tibble: 1 × 1
meanchildtot
<dbl>
1 4.69
*1st sum(rr = generating summary stats from dataset rr
*2 meanchiltot = giving name to sum stats
*na.rm=T ignores missing values
Can compute multiple var at once:
summarize(rr, meanchildtot = mean(childtot, na.rm=T),
medianchildtot=median(childtot, na.rm = T))
mean of chiltot taken by each value of dataset, store
under meanchildtot:
rr %>%
group_by(dataset) %>%
summarize(meanchildtot=mean(childtot,na.rm=T))
create unique combinations:A 2010, A 2015, B2010,
B2015
student.df %>%
group_by(school,graduationdate) %>%
summarize(mean.gpa=mean(gpa))
`summarise()` has grouped output by 'school'. You can
override using the `.groups` argument.
# A tibble: 4 × 3
# Groups: school [2]
school graduationdate mean.gpa
<chr> <dbl> <dbl>
1 A 2010 3.45
2 A 2015 2.9
3 B 2010 3.6
4 B 2015 1.8
Save output:
totchildbyyearcollege <- rr %>%
group_by(dataset, college) %>%
summarize(meanchildtot = mean(childtot, na.rm=T))
combination:
collegeprep <- rr %>%
filter(mother2==1, age>24, age<35) %>%
mutate(collegeprep = childeduc + childtravel) %>%
group_by(dataset, college) %>%
summarize(meancollegeprep=mean(collegeprep,
na.rm=T))
`summarise()` has grouped output by 'dataset'. You can
override using the `.groups` argument.
Filter wanted mother data, creat new var, group dta aby
dataset and college & summarize mean
Create histogram:
hist(pm_bycity$meanpm10,
xlab="Mean PM10",
ylab="Frequency",
main="Mean PM10 by City")
create multiple plots:
par(mfrow=c(1,2))
SINGLE ROW TWO COLUMNS (indicate with mfrow=)
#Smaller bins
hist(pm_bycity$meanpm10, xlab="Mean PM10",
ylab="Frequency", main="Mean PM10 by City",
breaks=20)
#Larger bins
hist(pm_bycity$meanpm10, xlab="Mean PM10",
ylab="Frequency", main="Mean PM10 by City",
breaks=4)
new var = T deciding date >/<auto_date:
pm_bycitybefore <- pm %>%
mutate(T=date-auto_date) %>%
filter(T<0) %>%
group_by(code_city) %>%
summarize(meanpm10 = mean(pm10, na.rm=TRUE))
pm_bycityafter <- pm %>%
mutate(T=date-auto_date) %>%
filter(T > 0) %>%
group_by(code_city) %>%
summarize(meanpm10 = mean(pm10, na.rm=TRUE))
comparing histograms:
par(mfrow=c(1,2))
#Plot histogram for before
hist(pm_bycitybefore$meanpm10, xlab="MeanPM10",
ylab="Frequency",
main="Before Automation")
#Plot histogram for after
hist(pm_bycityafter$meanpm10, xlab="MeanPM10",
ylab="Frequency",
main="After Automation")
same scale for both histograms:
#Create two panes for plots
par(mfrow=c(1,2))
#Plot histogram for before
hist(pm_bycitybefore$meanpm10, xlab="MeanPM10",
ylab="Frequency",
main="Before automation",
xlim=c(0,250), ylim=c(0,50))
lines(c(meanbefore, meanbefore), c(-10, 100),
lty=2, col="red")
#Plot histogram for after
hist(pm_bycityafter$meanpm10, xlab="MeanPM10",
ylab="Frequency",
main="After automation",
xlim=c(0,250), ylim=c(0,50))
lines(c(meanbefore, meanbefore), c(-10, 100),
lty=2, col="red")
boxplot function:
boxplot(pm_byday$meanpm10)
lines that extend from the box are referred to as
whiskers
whiskers – 1.5xIQR
IQR = Q3-Q1
*Horizontal, blue, with label box plot:
boxplot(pm_byday$meanpm10, xlab="Mean PM10",
main="Mean PM 10 by city",
col="blue", border="darkblue",
pch=16, horizontal=T)
separate box plots:
(1st add new var)
pm_byday$after <- pm_byday$T>0
boxplot(pm_byday$meanpm10 ~ pm_byday$after,
xlab="Mean PM10", ylab="After Automation",
main="Mean PM 10 by Day Before and After
Automation",
col="blue", border="darkblue",
pch=16, horizontal=T)
scatter plot:
plot(df$xvar, df$yvar)
*pch = .pch plot size
*lines() add line
EG) plot(pm_byday$T, pm_byday$meanpm10,
xlab="Days Relative to Automation",
ylab="Mean PM10",
main="Automation and Mean PM10",
pch=16, ylim=c(0,300))
lines(c(0,0), c(-10,250), col="red", lty=2)
text(0,260, "Automation", col="red")
save plot as pdf:
pdf("Automation.pdf")
dev.off() stop saving into pdf
library(lubridate)
today(), now()showsh date
*ymd(“2012-01-22”) convert string into date obj. can
also mdy/dmy/ymd
*can add days()/months()/years()
*portion – e.g year
year(ymd("2012-01-22"))
#> [1] 2012
*add
ymd("1960-01-01")+days(18628)
#> [1] "2011-01-01"
Plot mean amount of rain by month:
pm_bymonth <- pm %>%
mutate(rdate = ymd("1960-01-01") + days(date),
month = month(rdate)) %>%
group_by(month) %>%
summarize(meanrain = mean(rain, na.rm=TRUE))
plot(pm_bymonth$month, pm_bymonth$meanrain,
col="blue",
pch=16,
xlab="Date",
