Objective: I want to analyse hflight data with the objective of finding any relationship between Arrival Delay and Distance. I also want to find the most frequently delayed (arrival delay) carrier, the month in which most arrival delay happens, top 5 destinations and relationship between Departure Delay and Arrival Delay.
Select columns from original dataset and create a new dataset out of it.
# Select columns from hflight and create a new datset
newData <- subset(hflights,
select=c("Month", "DayOfWeek", "UniqueCarrier", "ArrDelay","DepDelay", "Dest", "Distance"),
na.rm = TRUE)
I have created another subsets by merging the Short and Long Distance DataSet. The new dataset has all the records for Delayed (Arrival Delay) Flights and will be helpful in overall analyses.The additional column FlightType can be used to compare data between Long Distance and Short Distance.
totalDelay <- rbind(sDistDelay, lDistDelay)
1.Lets analyse the Histogram Chart of Short Distance Flights and see which carrier is most frequently delayed on Arrival delay parameter.
ggplot(sDistDelay, aes(x=UniqueCarrier, y=frequency(UniqueCarrier))) + geom_bar(stat = "identity") +
labs(title="Arrival Delay for Short Distance Flights based on different Carriers", x="Carrier", y="Count")
From the above histogram chart, it is observed that the carrier which is most frequently delayed for Short distanced flights is ExpressJet Airlines (XE).
2.Similarly, lets analyse histogram for Long Distanced Flights
ggplot(lDistDelay, aes(x=UniqueCarrier, y=frequency(UniqueCarrier))) + geom_bar(stat = "identity") +
labs(title="Arrival Delay for Long Distance Flights based on different Carriers", x="Carrier", y="Count")
From the above histogram chart, it is observed that the carrier which is most frequently delayed for long distanced flights is CO.
3.Lets compare Arrival Delay of Long distance vs Short distance flights
qplot(ArrDelay, data = totalDelay, geom = "freqpoly", colour = FlightType, xlim =c(-20, 200)) +
labs(title="Long Distance vs Short Distance", x="Arrival Delay", y="Count")
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 780 rows containing non-finite values (stat_bin).
## Warning: Removed 6 rows containing missing values (geom_path).
From the graph we can see that there is not much difference between the distribution of Long Distance Arrival Delay vs short distance Arrival Delay, infact they both are very similar. Hence we can say that distance does not affect the Arrival Delay in any significant way.
4.Lets find out in which Month , most Arrival Delay happens for Short distanced flights.
sDistDelay$Month <- factor(sDistDelay$Month,levels=month.name)
sDistDelay = sDistDelay[order(sDistDelay$Month,decreasing=FALSE),]
# Which month has most delayed flights for Short Distance
ggplot(sDistDelay, aes(x=Month, y=frequency(Month))) + geom_bar(stat = "identity") + labs( x="Month", y="Count")
May is the month where most of the Arrival Delay happens for a short distanced flights
5.Lets find out in which Month , most Arrival Delay happens for Long distanced flights.
lDistDelay$Month <- factor(lDistDelay$Month,levels=month.name)
lDistDelay = lDistDelay[order(lDistDelay$Month,decreasing=FALSE),]
# Which month has most delayed flights for Long Distance
ggplot(lDistDelay, aes(x=Month, y=frequency(Month))) + geom_bar(stat = "identity") + labs( x="Month", y="Count")
June is the month where most of the Arrival Delay happens for a long distanced flights
6.Which Airline is responsible for most frequent Arrival Delay?
qplot(ArrDelay, data=totalDelay, geom='histogram', binwidth = 5, xlim =c(-25, 200), main='Arrival Delays by Airlines') +
facet_wrap(~UniqueCarrier) + labs(x = 'Arrival Delay', y = "Count")
## Warning: Removed 780 rows containing non-finite values (stat_bin).
XE followed by CO are the carriers responsible for Most Arrival Delay
7.In which month, most frequent Arrival Delay happens?
totalDelay$Month <- factor(totalDelay$Month,levels=month.name)
totalDelay = totalDelay[order(totalDelay$Month,decreasing=FALSE),]
# Histograms to compare Arrival Delay by Month
qplot(ArrDelay, data=totalDelay, geom='histogram', binwidth = 5, xlim =c(-25, 200), main='Arrival Delays by Month') +
facet_wrap(~Month) + labs(x = 'Arrival Delay', y = "Count")
## Warning: Removed 780 rows containing non-finite values (stat_bin).
June is the most in which most frequently Arrival Delay happens.
To analyse further, I concentrated my analyses on top 5 destination
# Get Top 5 Destination
topDest <- data.frame(table(newData$Dest))
topDest <- subset(topDest, Freq > 6000)
head(topDest)
## Var1 Freq
## 7 ATL 7886
## 29 DAL 9820
## 33 DFW 6653
## 60 LAX 6064
## 79 MSY 6823
# Filter Data by Top 5 Destination
topdest <- subset(newData, Dest == 'ATL' | Dest == 'DAL' | Dest == 'DFW' | Dest == 'LAX' |
Dest == 'MSY' )
8.Density chart comparison between top 5 destination showing Arrival Delay of differnt carriers
# Density chart comparison between top 5 destination showing Arrival Delay of differnt carriers
qplot(ArrDelay, data=topdest, geom='density', color=UniqueCarrier, xlim =c(-25, 150)) + facet_grid(Dest ~.)
## Warning: Removed 1495 rows containing non-finite values (stat_density).
ggplotly()
## Warning: Removed 1495 rows containing non-finite values (stat_density).
From the above density chart we can observe the most time bound and the most frequently delayed Airline for the top 5 destination from Houston. For example, for Atlanta, FL airline is most time bound and XE is most frequently delayed.
9.Which Top 5 destination has most frequency of delayed Arrival?
# Histograms to compare Arrival Delay by Top 5 Destination
topdest2 <- subset(totalDelay, Dest == 'ATL' | Dest == 'DAL' | Dest == 'DFW' | Dest == 'LAX' |
Dest == 'MSY' )
qplot(ArrDelay, data=topdest2, geom='histogram', binwidth = 5, xlim =c(0, 150), main='Arrival Delays by Destination') +
facet_wrap(~Dest) + labs(x = 'Arrival Delay', y = "Count")
## Warning: Removed 371 rows containing non-finite values (stat_bin).
From the above Histogram we can say that, Dallas(DAL) airport has most frequency of delayed arrival of flights.
10.Relationship between Arrival Delay and Departure Delay by Unique Carrier
# Scatter Chart to show relationship between Arrival Delay vs Departure Delay
b2 <- ggplot(newData, aes(x = DepDelay , y = ArrDelay))
b2 + geom_point(aes(color = UniqueCarrier)) + geom_smooth() +
labs(title="Departure vs Arrival Delay", x="Departure Delay", y="Arrival Delay")
## Warning: Removed 3622 rows containing non-finite values (stat_smooth).
## Warning: Removed 3622 rows containing missing values (geom_point).
From the above scatter chart we can conclude that there is a strong (alomost linear) positive correlation between Arrival Delay and Departure Delay