We being the tidying processing by transforming the messy data into a more useful format.

image

Screen shot of data
data.url <- file(paste(url,"Week%205/tidy%20data%20set.csv", sep = ""), open="r" )

flights <- read.csv(data.url, sep=",", header=TRUE, stringsAsFactors = FALSE)
flights <- flights[(rowSums(flights[3:7], na.rm = TRUE) > 0),] # removes blank row
flights$X[flights$X == ""] <- NA 
flights$X <- na.locf(flights$X) #copy forwards the column name for next blank cell
colnames(flights)[c(1,2)] <- c("Airline", "Flight_Status") 
flights$Flight_Status <- str_trim(flights$Flight_Status)

flights <- flights %>% 
           gather(Airline, Flight_Status, Los.Angeles:Seattle) 
colnames(flights)[c(3,4)] <- c("Location", "Counts")
kable(flights, align = "c")
Airline Flight_Status Location Counts
ALASKA on time Los.Angeles 497
ALASKA delayed Los.Angeles 62
AM WEST on time Los.Angeles 694
AM WEST delayed Los.Angeles 117
ALASKA on time Phoenix 221
ALASKA delayed Phoenix 12
AM WEST on time Phoenix 4840
AM WEST delayed Phoenix 415
ALASKA on time San.Diego 212
ALASKA delayed San.Diego 20
AM WEST on time San.Diego 383
AM WEST delayed San.Diego 65
ALASKA on time San.Francisco 503
ALASKA delayed San.Francisco 102
AM WEST on time San.Francisco 320
AM WEST delayed San.Francisco 129
ALASKA on time Seattle 1841
ALASKA delayed Seattle 305
AM WEST on time Seattle 201
AM WEST delayed Seattle 61

Now that the data is transformed into observations by row we can be graph the data to view the difference in flight data by delayed and the on time flights status. I set the colors for the graph using RColorBrewer using the Paired palette and using the blue and orange colors .

ggplot(flights, aes(x = Airline, y = Counts, fill = Flight_Status, group = Airline)) + 
  geom_bar(stat="identity", position=position_dodge(), colour="black") + 
  scale_fill_manual(values = rev(colors)) + 
  facet_grid(~Location) + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))

As we can see in the above graph there is a large disparity between airlines and the number of flights completed by location. To get a sense of which airlines has more delayed flights we can take the number of delayed flights as a percentage of all flights by each airline. Then we can graph this information as a boxplot to see how the airlines stack against each other for the percentage of flights delayed.

The below code with transform the data into a usable format for the boxplot.

flights_percent <- ddply(flights, ~Airline + Location, summarize, 'All Flights' = sum(Counts))
flights_delayed <- flights[(flights$Flight_Status == "delayed"),]
flights_delayed <- flights_delayed[c(1,3,4)]
colnames(flights_delayed)[c(3)] <- c("Delayed Counts")
flights_delayed <- merge(flights_percent, flights_delayed, by = c("Airline", "Location"), all = TRUE)
flights_delayed$Percent_delayed <- flights_delayed$`Delayed Counts` / flights_delayed$`All Flights`

The below graph uses violin bar charts to visualize the typical percent of delayed flights by each airline. We can see that Alaska Airlines appears to perform better for the percentage of delayed flights than AM West does for our given locations (Los.Angeles, Phoenix, San.Diego, San.Francisco, Seattle)

ggplot(flights_delayed, aes(x = Airline, y = Percent_delayed, fill = Airline)) + 
  scale_y_continuous(labels = percent, limits = c(0,.5)) + 
  geom_violin(trim = TRUE) + 
  scale_fill_manual(values = rev(colors)) + 
  geom_boxplot(width=0.1, fill = "white") +
  geom_jitter(width=0.1) + 
  theme_minimal() +
  theme(legend.position = "top") + 
  ylab("Percent of Flights Delayed\n")