HW4

Step 1: Read CSV file from web, review data

raw.flights <- read.csv("https://raw.githubusercontent.com/aliceafriedman/DATA607_HW4/master/a_flights.csv", sep=",", header=TRUE, stringsAsFactors=FALSE)

glimpse(raw.flights)

## Observations: 5
## Variables: 7
## $ X             <chr> "Alasaka", "", "", "AM WEST", ""
## $ X.1           <chr> "On time", "Delayed", "", "On time", "Delayed"
## $ Los.Angeles   <int> 497, 62, NA, 694, 117
## $ Phoenix       <int> 221, 12, NA, 4840, 415
## $ San.Diego     <int> 212, 20, NA, 383, 65
## $ San.Francisco <int> 503, 102, NA, 320, 129
## $ Seattle       <int> 1841, 305, NA, 201, 61

Step 2: Add missing column names, remove blank row

# Add column names where missing, remove blank row
flights <- raw.flights %>% 
  rename("Airline"=X, "Status"=X.1) %>% 
  na.omit() %>%
  glimpse()

## Observations: 4
## Variables: 7
## $ Airline       <chr> "Alasaka", "", "AM WEST", ""
## $ Status        <chr> "On time", "Delayed", "On time", "Delayed"
## $ Los.Angeles   <int> 497, 62, 694, 117
## $ Phoenix       <int> 221, 12, 4840, 415
## $ San.Diego     <int> 212, 20, 383, 65
## $ San.Francisco <int> 503, 102, 320, 129
## $ Seattle       <int> 1841, 305, 201, 61

Step 3: Add missing data to “Airline” variable

#Note: This is an easy place to introduce errors into the dataset! 
#This would also not work wel for a larger data set.
flights$Airline[2] <- "Alaska"
flights$Airline[4] <- "AM WEST"
glimpse(flights)

## Observations: 4
## Variables: 7
## $ Airline       <chr> "Alasaka", "Alaska", "AM WEST", "AM WEST"
## $ Status        <chr> "On time", "Delayed", "On time", "Delayed"
## $ Los.Angeles   <int> 497, 62, 694, 117
## $ Phoenix       <int> 221, 12, 4840, 415
## $ San.Diego     <int> 212, 20, 383, 65
## $ San.Francisco <int> 503, 102, 320, 129
## $ Seattle       <int> 1841, 305, 201, 61

The above is an easy way to introduce errors into the data*, and won’t work well in larger dataset. Is there a better way to do it?

i <- 2
while(i <= length(flights$Airline)) {
  if(flights$Status[i]=="Delayed"){
    flights$Airline[i] <- flights$Airline[i-1]
    }
  i <- i+2
}
glimpse(flights)

## Observations: 4
## Variables: 7
## $ Airline       <chr> "Alasaka", "Alasaka", "AM WEST", "AM WEST"
## $ Status        <chr> "On time", "Delayed", "On time", "Delayed"
## $ Los.Angeles   <int> 497, 62, 694, 117
## $ Phoenix       <int> 221, 12, 4840, 415
## $ San.Diego     <int> 212, 20, 383, 65
## $ San.Francisco <int> 503, 102, 320, 129
## $ Seattle       <int> 1841, 305, 201, 61

*Note, in fact, that the data in flights$Airline doesn’t match! There is an error in the CSV, which we will correct below.

flights$Airline <- gsub("Alasaka", "Alaska", flights$Airline)

Step 4: Gather flights data to analyze delay between airlines

flights.long <- flights %>%
  gather(key = city, value = count, Los.Angeles, Phoenix, San.Diego, San.Francisco, Seattle) %>%
  glimpse()

## Observations: 20
## Variables: 4
## $ Airline <chr> "Alaska", "Alaska", "AM WEST", "AM WEST", "Alaska", "A...
## $ Status  <chr> "On time", "Delayed", "On time", "Delayed", "On time",...
## $ city    <chr> "Los.Angeles", "Los.Angeles", "Los.Angeles", "Los.Ange...
## $ count   <int> 497, 62, 694, 117, 221, 12, 4840, 415, 212, 20, 383, 6...

Compare overall delay records

First, we will compare the overall record for each airline. Overall, AM WEST flys substantially more flights overall, and has more flights with delays, but as a percentage of all flights does slightly better than Alaska. This seems like AM WEST is your better bet to avoid delays!

compare.all <- flights.long %>% 
  select(-city) %>% 
  group_by(Airline) %>% 
  mutate(All.Flights = sum(count)) %>% 
  group_by(Airline, Status) %>% 
  mutate(Flights = sum(count), Percent.By.Status = Flights/All.Flights*100) %>%
  group_by(Airline, Status, Flights, Percent.By.Status) %>% 
  summarise()

#Graph number of flights by status, airline
ggplot(compare.all, aes(x=Status, y=Flights, fill=Airline))+geom_col(position = "Dodge")+
  ggtitle("Total Number of Flights by Airline, Status")

#Graph proportion of all flights by status, airline
compare.all %>% filter(Status=="Delayed") %>% 
  ggplot(aes(x=Airline, y=Percent.By.Status, fill=Airline))+geom_col()+
  ggtitle("Frequency of Delayed Flights, by Airline")+
  ylab("% of All Flights Delayed")

Compare Delay Frequency by City

Just to be sure, we will compare the frequency of delayed flights arriving in each city. Per the data below, your best bet is actually always to take Alaska, unless you are flying equally to all cities!

Looks like the high number of delayed flights going into Seattle may be skewing the overall numbers.

compare.by.city <- flights.long %>% 
  group_by(Airline, city) %>% 
  mutate(
    City.Flights = sum(count), 
    Status.Freq.By.City = count/City.Flights*100, 
    Delayed.Flights = count) %>%
  filter(Status == "Delayed") %>%
  select(-Status, -City.Flights, -count) %>% 
  arrange(city) %>% 
  as_tibble()

compare.by.city

## # A tibble: 10 x 4
## # Groups:   Airline, city [10]
##    Airline city          Status.Freq.By.City Delayed.Flights
##    <chr>   <chr>                       <dbl>           <int>
##  1 Alaska  Los.Angeles                 11.1               62
##  2 AM WEST Los.Angeles                 14.4              117
##  3 Alaska  Phoenix                      5.15              12
##  4 AM WEST Phoenix                      7.90             415
##  5 Alaska  San.Diego                    8.62              20
##  6 AM WEST San.Diego                   14.5               65
##  7 Alaska  San.Francisco               16.9              102
##  8 AM WEST San.Francisco               28.7              129
##  9 Alaska  Seattle                     14.2              305
## 10 AM WEST Seattle                     23.3               61

ggplot(compare.by.city, 
       aes(x=city, y=Status.Freq.By.City, fill=Airline))+
  geom_col(position = "dodge")+
  ggtitle("Percent Flights Delayed by City")+
  xlab("City")+
  ylab("% Delayed")