Solution to Sample Midterm Stat 480

Work on questions in R, make sure to keep a copy of your R code - you will be asked to submit this script at the end.

Data on all flights in and out of Des Moines (DSM) for October 2008 are available at http://www.hofroe.net/data/dsm-flights.csv.
See http://stat-computing.org/dataexpo/2009/the-data.html for a description of the variables.

1. Load the flights data into R.
   Determine, which flight was delayed the most on arrival. Report its row number, where it started, and by how much the flight was delayed on departure.

flights <- read.csv("http://www.hofroe.net/data/dsm-flights.csv")
which.max(flights$ArrDelay)

## [1] 1516

flights[which.max(flights$ArrDelay), c("Origin", "DepDelay")]

##      Origin DepDelay
## 1516    DSM      614

2. Bring the variable ‘Day’ into the correct order, starting with ‘Monday’.

summary(flights$Day)

##    Friday    Monday  Saturday    Sunday  Thursday   Tuesday Wednesday 
##       456       368       248       336       460       360       459

days <- levels(flights$Day)
flights$Day <- factor(flights$Day, levels=days[c(2,6,7,5,1,3,4)])
summary(flights$Day)

##    Monday   Tuesday Wednesday  Thursday    Friday  Saturday    Sunday 
##       368       360       459       460       456       248       336

3. Create a new variable called ‘Weekend’ which has value TRUE for Saturdays and Sundays and FALSE otherwise.

# create new variable Weekend
flights$Weekend <- flights$Day %in% c("Saturday", "Sunday")

summary(flights$Weekend)

##    Mode   FALSE    TRUE 
## logical    2103     584

4. Determine how many flights arrived in Des Moines on average each day of the week.

# idea 1:
table(subset(flights, Dest=="DSM")$Day) # overall number of flights by day of week

## 
##    Monday   Tuesday Wednesday  Thursday    Friday  Saturday    Sunday 
##       184       180       230       230       228       124       168

# problem: how many Mondays, Tuesdays, are there in October 2008?

require(lubridate)
octs <- data.frame( date = ymd(paste("2008/10/",1:31, sep="")))
octs$day = wday(octs$date, label=TRUE)
table(octs$day)

## 
## Sun Mon Tue Wed Thu Fri Sat 
##   4   4   4   5   5   5   4

table(subset(flights, Dest=="DSM")$Day)/c(4,4,5,5,5,4,4)

## 
##    Monday   Tuesday Wednesday  Thursday    Friday  Saturday    Sunday 
##      46.0      45.0      46.0      46.0      45.6      31.0      42.0

# idea 2: 
require(dplyr)
require(lubridate)
flights %>% filter(Dest == "DSM") %>% group_by(DayofMonth) %>% summarise(
  day = Day[1],
  n = n()
) %>% group_by(day) %>% summarize(avg = mean(n), n = n())

## # A tibble: 7 x 3
##   day         avg     n
##   <fct>     <dbl> <int>
## 1 Monday     46       4
## 2 Tuesday    45       4
## 3 Wednesday  46       5
## 4 Thursday   46       5
## 5 Friday     45.6     5
## 6 Saturday   31       4
## 7 Sunday     42       4

6. How many flights were scheduled to go to Denver (DEN)? What percentage of flights goes to Denver?

nrow(subset(flights, Dest=='DEN'))

## [1] 145

nrow(subset(flights, Dest=='DEN'))/nrow(subset(flights, Dest != 'DSM'))*100

## [1] 10.79672

7. Where do most flights arriving in Des Moines come from? (use IATA code)

flights %>% 
  filter(Origin != "DSM") %>% 
  group_by(Origin) %>% 
  summarise(n = n()) %>% 
  arrange(desc(n))

## # A tibble: 13 x 2
##    Origin     n
##    <fct>  <int>
##  1 ORD      379
##  2 DFW      178
##  3 DEN      119
##  4 ATL       89
##  5 MSP       89
##  6 MKE       85
##  7 CVG       80
##  8 MEM       62
##  9 DTW       58
## 10 IAH       58
## 11 LGA       58
## 12 PHX       58
## 13 DCA       31

sort(table(flights$Origin), decreasing=T)[2]

## ORD 
## 379

8. Plot boxplots of arrival delays by originating airports. Order boxplots according to increasing median arrival delay.

library(ggplot2)
flights %>% filter(Dest =="DSM") %>%
ggplot( aes(x = reorder(factor(Origin), ArrDelay, na.rm=T), y = ArrDelay)) + geom_boxplot()

9. Using dplyr, determine for flights leaving DSM for each hour of the day

• the number of flights scheduled to leave,
• the percentage of flights departing late (FAA considers flights late, if they are 15 minutes or more behind the scheduled time),
• the average departure delay, and
• the top 3 destinations.

Draw a scatterplot of average departure delay by scheduled hour of departure. Color points by top destination, adjust point size to reflect the number of flights for each hour.

dep.summary <- flights %>% 
  filter(Origin == 'DSM') %>%
  mutate(hour = CRSDepTime%/%100) %>% 
  group_by(hour, Dest) %>%
  mutate(test = n()) %>% 
  group_by(hour) %>% 
  mutate(rank = dense_rank(desc(test))) %>% 
  summarise(
    count = n(),
    pct.delayed = sum(DepDelay>15, na.rm=TRUE)/n()*100,
    avg.delay = mean(DepDelay, na.rm=T),
    top.Dest.1 = Dest[which.max(test)],
    top.Dest.2 = first(Dest[which(rank == 2)]),
    top.Dest.3=first(Dest[which(rank == 3)]))


dep.summary %>% 
  ggplot(aes(x = hour, avg.delay, colour = top.Dest.1, size = count)) + geom_point()