R Bridge Final Project

Logan Thomson

Loading Libraries

library(hflights)
library(ggplot2)
library(dplyr)

## 
## Attaching package: 'dplyr'
## 
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## 
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

Transforming the Data

Change Days of Week to Character Values

Days of the week were numerical, so instead of having to remember what day of the week corresponded to which number, I changed the DayOfWeek values to the appropriate character values

hflights$DayOfWeek <- gsub(1, 'Monday', hflights$DayOfWeek)
hflights$DayOfWeek <- gsub(2, 'Tuesday', hflights$DayOfWeek)
hflights$DayOfWeek <- gsub(3, 'Wednesday', hflights$DayOfWeek)
hflights$DayOfWeek <- gsub(4, 'Thursday', hflights$DayOfWeek)
hflights$DayOfWeek <- gsub(5, 'Friday', hflights$DayOfWeek)
hflights$DayOfWeek <- gsub(6, 'Saturday', hflights$DayOfWeek)
hflights$DayOfWeek <- gsub(7, 'Sunday', hflights$DayOfWeek)

Combine Date Columns

flight_date <- paste(hflights$Month, hflights$DayofMonth, hflights$Year, sep="/")

hflights$Date <- flight_date
hflights$Date <- as.Date(flight_date, "%m/%d/%Y")

Now that I have some new data, and some columns that I’m not interested in, I combined the data into a new data frame.

myflights <- subset(hflights, select=c(Date, DayOfWeek, UniqueCarrier, FlightNum, TailNum, TaxiIn, TaxiOut, AirTime, ArrDelay, 
                                       DepDelay, ActualElapsedTime, Origin, Dest, Distance))

Add a Distance/Time Value

I created a combined field by dividing the total distance of the flight by the “Actual Elapsed Time” field. This gave me a number I could use to convey the range of time it took to fly the same distance.

dis_per_min <- (myflights$Distance / myflights$ActualElapsedTime)
myflights$mi_per_min <- dis_per_min

Here is a preview of my new data frame:

head(myflights)

##            Date DayOfWeek UniqueCarrier FlightNum TailNum TaxiIn TaxiOut
## 5424 2011-01-01  Saturday            AA       428  N576AA      7      13
## 5425 2011-01-02    Sunday            AA       428  N557AA      6       9
## 5426 2011-01-03    Monday            AA       428  N541AA      5      17
## 5427 2011-01-04   Tuesday            AA       428  N403AA      9      22
## 5428 2011-01-05 Wednesday            AA       428  N492AA      9       9
## 5429 2011-01-06  Thursday            AA       428  N262AA      6      13
##      AirTime ArrDelay DepDelay ActualElapsedTime Origin Dest Distance
## 5424      40      -10        0                60    IAH  DFW      224
## 5425      45       -9        1                60    IAH  DFW      224
## 5426      48       -8       -8                70    IAH  DFW      224
## 5427      39        3        3                70    IAH  DFW      224
## 5428      44       -3        5                62    IAH  DFW      224
## 5429      45       -7       -1                64    IAH  DFW      224
##      mi_per_min
## 5424   3.733333
## 5425   3.733333
## 5426   3.200000
## 5427   3.200000
## 5428   3.612903
## 5429   3.500000

Flight Times and Delays by Day of Week

Plotting Flight Time, Distance, and Day of Week

I wanted to see if there was any pattern to the distance traveled, the flight time, and the day of the week.

qplot(Distance, ActualElapsedTime, color=DayOfWeek, data=myflights, main='Total Flight Time by Distance')

## Warning: Removed 3622 rows containing missing values (geom_point).

What about departure delays by day of the week? Looks like the days with greater amounts of delays simply have more flights.

qplot(DepDelay, data=myflights, geom='histogram', binwidth = 5, xlim =c(-25, 200), main='Departure Delays by Day of Week') + facet_wrap(~DayOfWeek)

I decided to work with a smaller data set, so I looked at the airlines with the most flights, and the destinations with the most flights.

ggplot(data=myflights) + geom_bar(aes(x=UniqueCarrier), bin=5) + ggtitle('Flights by Airline')

Looks like there are a lot more destinations to work with, so I will reduce the number to the destinations with the most flights.

dests <- ggplot(data=myflights) + geom_bar(aes(x=Dest))
dests + theme(axis.text.x=element_text(angle=-90)) + ggtitle('Flights by Destination')

Using The Top Destinations Only

I could have used dplyr to get this information, but I wanted to get familiar with various ways of selecting data. Here, I use table to select the Destination column, and turn that into a data frame. I can then subset that to destinations with 5000 flights or more.

top_dests <- data.frame(table(myflights$Dest))
top_dests <- subset(top_dests, Freq > 5000)
head(top_dests)

##    Var1 Freq
## 7   ATL 7886
## 8   AUS 5022
## 29  DAL 9820
## 32  DEN 5920
## 33  DFW 6653
## 60  LAX 6064

Now that I know the top destinations, I will create a dataframe with the flights information, but only for those selected cities:

most_dest <- filter(myflights, Dest == 'ATL' | Dest == 'AUS' | Dest == 'DAL' | Dest == 'DEN' | Dest == 'DFW' | Dest == 'LAX' | 
                      Dest == 'MSY' | Dest == 'ORD' | Dest == 'PHX')

Plotting Data

Destinations with highest frequency of flights, range of distance covered per minute (includes taxi time, flight, delays)

Destinations with miles per minute, by carrier - who is the best performer per destination?

qplot(mi_per_min, data=most_dest, geom='density', color=UniqueCarrier) + facet_grid(Dest ~.)