Sie sind auf Seite 1von 29

American Airline

On-Time performance analysis

Team Member
Jingming
Juyang
Xinyan
Anton

Executive Summary
To long-term success for American Airline, improving on-time performance is one of the key
indicators for overall corporate responsibility performance. The intention of this project is to
analysis data of American Airline On-Time Arrival performance to support Mission Objective
pertaining to Customer loyalty, and our stated objective of dependable travel options. The data is
from August 2014 to July 2015 which includes flight number, particular routes, airports, the
cause of the delay. The aim is to find the trends or patterns to make a change to our operations.
Due to the high volume of data, a database is set up, and the analysis was carried out using SAS
9.2 software. We defined flights delay is more than 15 minutes of arrival.
The report focuses on three main pieces of analysis to fulfill the goal. First, the factors that lead
to high delay time need to be understood. Using procedure means and frequency distribution
applied to a large historical dataset to infer the causes of delay of airlines. The report indicates
that carrier delay and late arriving delay are main reasons for late arrivals, and late craft delay
caused higher delay time than carrier delay on the average.
Second, the report characterizes the top 20 offenders of particular planes, particular routes, and
destination within the subset of data to obtain the insights of late arrivals. By using frequency
procedure to examine the flights tail number on delays. The results indicate that ten flights
number are from the same manufacturer within all the delay flights for American Airlines. DFW,
ORD, MIA are the top 3 airports with the highest delay rate in both origins and destinations.
Finally, using the means procedure to compare our principal competitors (Delta, Southwest,
United, and JetBlue) by calculating the average delay times, on time times and delay rate. From
the results, American Airline does not have the good performance among the competitors
because of the highest average delay time and the highest standard division. Hence, American
Airline does not have a stable on-time performance.
There are some recommendations to lend to Customer Loyalty Strategy based on analyzing the
on-time performance data. The recommendations will be useful to succeed in the strategic plan
and increase overall corporate responsibility performance.

Introduction & Background What Is the Strategic Basis


American Airline (AA) is a major airline of the world, it is the biggest airline measured by fleet
size, revenue, and scheduled passenger-kilometers flown, and second largest measured by
number of destinations served. AA operates an extensive international and domestic network in
more than 50 countries, nearly 6,700 flights take off every day to nearly 350 destinations.
However, AA ranked 77 in the list of the World's Top 100 Airlines in 2016 (Skytrax), voted by
airline customers around the world, and its main competitors JetBlue ranked 53, United Airlines
ranked 68, Delta ranked 35, and Southwest ranked 66. AA has a relatively low customer
satisfaction compared with its competitors, and delay is an important cause that lead to the
disadvantage in ranking. Last year, AA had the highest number of flights delayed or cancelled
altogether, which has a negative influence on AAs reputation, and subsequently affected its
business.
In order to improve the customer satisfaction and gain customer loyalty, AA should begin with
analyzing its on-time performance and improve the customer travel experience according to the
analysis, which also corresponds with five tenets of Flight Plan 2020:
Earn Customer Loyalty: on-time performance is an important factor that affect customer
loyalty, and improving customer travel experience can help gaining customer loyalty;
Strengthen and Defend our Global Network: by analyzing ontime performance, we can know
which network alignment or carrier has a good performance and which does not, so that make
adjustments on partnership.
Be a Good Place for Good People: improving on-time performance can improve the customer
impression, reduce complaints and build a responsible company image.
Invest Wisely: by analyzing ontime performance, we can know which models of plane have
better performance which models have not, thus invest wisely on good ones and reduce invest of
bad performance models.

Fly Profitably: we must be competitive in everything including costs, product, revenue, service,
and schedule to keep our competitively and improve fly profitably.
Analyzing on-time performance corresponds to various aspects of our strategic plan, and help to
strengthening our overall corporate responsibility performance.

Data/Data Collection Plan


Objective:
Analyzing the on-time performance of AA, and compared its on-time performance with its
competitors.
Measures:
12 months flight operations information (including flight date, airline ID, tail number, origin,
destination, arrival delay, distance, and for causes of delay: carrier, weather, NAS, security and
late aircraft), and details of a particular aircraft (from FAA Registration).
Type of data:
Both quantitative and qualitative data are involved in the research, in details, we have discrete
data, attribute data, continuous data, binomial data, and nominal data
Data sources:
Existing 12 months archive of flight operations information, FAA Registration
Methods:
First, using the existing 12 months archive and SAS Program for data scrubbing and reducing,
and summarize and visualize the data that can indicate the correlation between delay and the four
causes of delay, delay and seasons, and compare AA and its four competitors numerical delay
information. Second, using FAA Registration Number lookup website to get the detail of a
particular aircraft.

Data Issues/Approach for Scrubbing & Reducing


We use SAS Program for data scrubbing & reducing.
Firstly, our group input all twelve files by a correct path.
After entering the path, we start to input variables,
Characters variables:
FL_DATE, AIRLINE_ID, TAIL_NUM, ORIGIN, DEST
Numbers variables:
ARR_DELAY, DISTANCE, CARRIER_DELAY, WEATHER_DELAY, NAS_DELAY,
SECURITY_DELAY, LATE_AIRCRAFT_DELAY
And we use (FIRSTOBS=2) function to ask SAS start observe data from the second row. Since
the first row are the categories of the data, which does not contain any data in it.
Then we use Set function to tell SAS we need to use all 12 months data to compare
Using (OBS=10) to give us a table with the first 10 records.
Use if function (IF airline=19805) to find out we are now looking at all data related to the Airline
code 19805, which is American Airline.
The next step we use is a proc means data=all_flight; to find mean procedure for these five
causes var cause_carrier, cause_WEATHER,cause_nas cause_SECURITY , cause_late_arr. And
this will give us a table shown as below stating variable, mean, and standard deviation, minimum
and maximum of the delay minutes of American Airline.

By using the sort data function to sort data by descending tail and the function of arr_delay ge 16
to limit the effective data range greater or equal than 16 minutes.
Plus the command (OBS=20) to show the top 20 among the data, it will generate a table shown
as below.

The table is showing the frequency, percent, cumulative frequency, and cumulative percent of the
top 20 top 20 flights tail numbers
Following step will be knowing the American Airline top 20 flights with the highest arriving
delay. Using sort function to sort data=all_flights by descending arr_delay with (OBS=20).

Then we use similar method to generate five main reason causing arriving delay, which are
Carrier Delay, Weather Delay, National Airspace System Delay, Security Delay, and Late Aircraft
Delay.
And we also want to know the top10 delay flights origin are. We use proc freq data=all_flights,
tables Origin to generate a table to sort the frequency of origin by descending origin, plus
arr_delay ge 16 and (OBS=10). The outcome is shown as below.

We used the same way to find the top10 delay flights destination.
After knowing top10 delay flights origin and destination, we decided to use mean table as
explained above, to find average delay information for American Airline and its four main
competitors. By using if function, we will be able to scrub information using the same method
but different airline code. The airline code is listed below:

if American Airline=19805;
if jetblue=20409
if United=19977
if Southwest=19393
if Delta=19790
This is the mean procedure table for American Airline.

We use a proc format, value latefmt low - <16 = 'On Time', and 16 - high = 'late' to distinguish
delay more than 15 mins will count as delay while within 15 mins will count as arrive on time.
With the help of data mod_all_flights;
set all_flights, Late_Cat=put(arr_delay, latefmt.)
and proc freq data=all_flights; tables arr_delay/nocum nopercent;
format arr_delay latefmt., plus if American airline= 19805. It will generate a frequency
distribution table showing the on time and delay frequency.

By using different airline code will be able to generate same table for different airline.
if jetblue=20409
if United=19977
if Southwest=19393
if Delta=19790

In order to find the frequency of severe delays of American Airline caused by Carrier, we set up a
Proc format; using
value latecarrier 0 - <100 = '100 minuts late'
100 - <500 = '100 to 500 Minutes Late'
500 - <1000 = '500 to 1000 Minutes Late'
1000 - high = '1000 minutes or More late';
proc freq of tables cause_carrier/nocum nopercent; format cause_carrier latecarrier.; to create a
table shown as below.

Lastly, sorting the data seasonal (spring, summer, fall and winter) can help us understand more
on the difference of delay flights by using the same measure of on time and delay frequency, five
causes of delaying frequency and delay rate to show the relationship between season and delay.

Analytical Approach
Our goal is to identify the patterns or trends of American Airlines delay data. And the delay
causes can be breakdown into five main causes which are Carrier Delay, Weather Delay, National
Airspace System Delay, Security Delay, and Late Aircraft Delay. By observing the data shown
below, we will be able to distinguish which one is the main causes for the delay.

(Fig.1)
Then, we consider seasonal would also be a reason causing delay, like typhoon in summer and
snowstorm in winter. Therefore, we have put into table shown as below to show the late
frequency in different season.
Fig.2 is AAs summer delay frequency while Fig.3 is AAs fall delay frequency

(Fig.2)

(Fig.3)

In this case, it is clearly shows that the delay frequency in summer is more than fall, then AA will
be able to make adjustment and improvement under different weather condition.
Besides comparing to the industry average, we also want to know how American Airline did
verse four of their main competitors Delta, Southwest, United and JetBlue.

On time
frequency

American
Airline

JetBlue

United

Southwest

Delta

441063

202576

385475

961362

727649

51263

109774

247026

115251

Delay frequency 115781

(Fig.4)
Fig.4 is showing the on time frequency and delay frequency of AA and its four main
competitors. By comparing the data, we will be able to understand how well American Airline
is doing.
Within the subset of data for which we have control (Carrier Delay and Late Aircraft Delay), we
have characterized the top 20 delay flight in American Airline according to the data.

(Fig.5)
Fig.5 shows a clear vision of we have characterize the top 20 offenders by listing their particular
planes (N number), a particular route, and airports.
Finally, we use the FAA Registration Number lookup website
(http://registry.faa.gov/aircraftinquiry/nnum_inquiry.aspx) to find out the information about a
specific plane.

(Fig.6)

Fig.6 shows that by using N number, it can lead us to the information of a specific plane, for
example when the plane was manufactured, what is the engine modeletc.

Analysis
First Factor: cause_carrier and cause_late_arr.
After carefully imported all the data, the team first want to analyze which factor is the dominant
reason for causing American Airline flights delay.

Using Procedure Means in SAS the team conclude that Carrier delay and late arriving delay are
dominating late arrivals. These two factors have the highest mean among all the five factors.
Carrier delay caused the most severe delay: 2402 minutes for flight departing from LAX heading
to DFW. But on average, late craft delay caused higher delay time than carrier delay.

But that doesnt mean other factors are not significant. The team decided to run a correlation
coefficient between the five factors and arriving delay to find the P values and use P value to
determine statistical significance in a hypothesis test.

As the team observed from the table, except cause_security, all other four factors have a P-value
, which indicates that the correlation is statistically significant. Also cause_carrier and
cause_late_arr have the highest value So the team conclude that cause_carrier and cause_late_arr
are the factors that influence delay.

Second factor: day of week.


After finding the dominant reasons, the team wants to look deeper to find out if there are any
patterns for those delay flights. So proc format was used to find out what is the delay rate for
each day of a week.

AA on time rate in each day of a week


Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday

78.30%
75.80%
79.52%
80.80%
78.21%
77.85%
81.64%

The team found out that for AAs flight, on time rate is the highest on Wednesday and Saturday,
the lowest on Monday. So we think day of week is another factor that influence AAs flight delay.

Third factor: Tail Number


Under airline=19805 and arr_delay ge 16, the top 20 flights tail numbers are as listed:

From the graph we found:

N4UBAA,N4YBAA,N4XWAA, N4WLAA,N4WKAA, N4XPAA,N4XSAA are not valid code,


N202AA is CESSNA 421C
N425AA is PIPER PA-28-180
the rest 10 flights all have the same Manufacture Name: Mcdonnel-Douglas MD-82)

Here are the American Airline top 20 flights with the highest arriving delay:

And Under airline-19805 and arr_delay ge 16, top 20 flights with highest carrier-delay:

And Under airline-19805 and arr_delay ge 16, top 20 flights with highest late_craft_delay:

Among all the delay flights from American Airline, 10 of the top 20 flights tail number comes
from the same manufacturer with the same aircraft type which is: Mcdonnel-Douglas type DC-982(MD-82).
From all the top 20 flights with highest arriving delay the team found that 4 out of 20 flights are
from Mcdonnel-Douglas model MD-82. 4 out of 20 are from Mcdonnel-Douglas model MD-83.
The rest are with invalid N number.
From all the top 20 flights with highest carrier-delay the team found that 5 out of 20 flights are
from Mcdonnel-Douglas model MD-82. 4 out of 20 are from Mcdonnel-Douglas model MD-83.
Flights with highest late craft delay dont have a clear pattern. Except all flights with invalid N
number, the rest came from three different manufacturers:
1. SIKORSKY S-76A
2. Mcdonnel-Douglas DC-9-82 (MD-82)
3. Boeing 767-323

Fourth factor: Airports (Origin and Destination).

whats more, when the team look at the top 20 offenders with highest arriving delay they found
that a lot of them were either departing from Dallas/Fort Worth International Airport or heading
to Dallas/Fort Worth International Airport(DFW). So next, an analysis based on the top 10
airports with the highest number of delay flights was designed.

Under airline=19805 and arr_delay ge 16, the top10 delay flights Origin are:

Under airline=19805 and arr_delay ge 16, the top 10 delay flights Destination are:

So from the top 10 origins and destinations with the highest rate of delay the team sees that DFW
is the top 1 in both tables. It has the highest percentage compare to other airports, followed by
O'Hare International Airport and Miami International Airport.
The team then run a Chi-Square to see is the high probability that those top 5 airports are causing
delay only by chance or there is more to dig in.

From the Chi-Squares Probability value, the team conclude that the distributions are different
and that there is a very low probability that the differences are due to chance alone. So the team
conclude that airport is another factor that influence delay.

Fifth Factor: Season


Then the team decided to see what is the patterns in four seasons

AA On time rate in Four seasons


Spring
Summer

79.98%
78.49%

Fall
79.38%
Winter
77.49%
In order to know whether these factors are statistically significant, a simple linear regression
model using arr_delay as dependent variable and day of week, season, cause_carrier,
cause_late_arr:

After carefully observed the Pr value in the result table the team conclude that all four variables
are statistically significant.

Next, we want to compare American Airline with its competitors. We calculated the average
delay times for each of the companies.

Airlines delay rate are listed:


AA

JetBlue

United Air

Southwest

Delta

On time times

441062

202576

385475

961362

727649

Delay times

115781

51263

109774

247026

115251

Average Delay
time

62.21

59.72

62.08

47.60

60.55

Maximum
delay time

2444

952

1294

825

1274

Delay rate

20.79%

20.20%

22.17%

20.44%

13.67%

Delay rate (Frequency Missing excluded) of AA is 20.79%, rates of JetBlue, United Air,
Southwest and Delta are 20.20%, 22.17%, 20.44% and 13.67% as listed, United Air is the first
place and Delta is the least. American Airline is doing a bad job comparing to its competitors. It
has the highest average delay time and it also has the highest std dev. which indicate that AA has
an unstable performance. Also, American Airline has the second highest delay rate.
We then want to look at for American Airlines competitors, what is the dominant reason for
causing their delays.
AA

JetBlue

United Air

Southwest

Delta

Carrier

60350

32707

70526

161971

63099

Weather

8908

1929

7378

9759

10998

NAS

67661

32300

58778

108388

68957

Security

350

358

17

564

53

Late aircraft

58519

26939

53829

184007

50398

Notes

1)biggest
mean
attributed to
late craft
delay

1)biggest
mean
attributed to
late aircraft

1)biggest
mean
attributed to
late aircraft

1)biggest
mean
attributed to
late aircraft

1)biggest
mean
attributed
to carrier

2)max
belong to
carrier (942
mins)

2)max
belong to
carrier
(1088 mins)

2)max
belong to
carrier (825
mins)

2)max
belong to
carrier
(1274
mins)

2)max
belong to
carrier
(2402 mins)

From the table we found that:

NAS caused most of the delays of aa and Delta, doubled the delay times of JetBlue, while carrier
caused most delays of United Air and JetBlue, Late aircraft caused most delays of Southwest;
Security caused few delay for United Air, late aircraft caused far more delay for Southwest
compared with other airline companies;
For AA, biggest mean attributed to late craft delay, (which means late craft is the most important
factor considering both delay durations and times), so does Delta;
Biggest mean attributed to late aircraft for JetBlue, United Air and Southwest;
Carrier caused longest delay for all 5 airline companies, the highest is AA (2440 mins);

Conclusions & Recommendations

Conclusions
In this report, we have investigated the factors influencing flight on-time performance, including
airplane model, aircraft age, destinations of airport, weather, Carrier, National airspace system,
security and late aircraft delay. We use SAS to run a set of historical flights over one year to
infer the trends and patterns of on-time performance with improving flight delay decisions.
Our results indicate arrive delay is highly related to carrier delay and late arrive delay which
have the highest mean with all the five factors. Particularly, the procedure means table gives the
maximum minutes of aircraft delay is carrier delay caused. Based on the mean derived from the
raw data, the later craft delay has higher delay time than the carrier delay. These means we need
to pay more attention to late carrier delay and we can reduce delay time by improvements.
We evaluated the TOP 20 flights tail numbers with the highest arriving delay and found ten
flights tails come from the same Manufacture with same aircraft type which manufacturer's name
is McDonnell-Douglas model MD-82. Meanwhile, we also found most routes with highest
arriving delay are either departing from DFW, ORD, and MIA or landing to DFW, ORD and
MIA. These three airports indicate arrival delay are related to the airport.
We then compared American Airline with principal competitors, including Delta, Southwest,
United, and JetBlue. The results show American Airline Delay rate is 20.79% which is not the
lowest delay rate within the competitors. Most of the delay causes are National Airspace System
which has the highest delay cause. National Airspace System is the cause which we cannot in
control so we need to decrease the delay rate of other reasons which we can control.
The seasonal variables indicate the seasons when the flights are scheduled. Observe from
seasonal delay flights table, and the seasonal variables show the seasons when the flights are
scheduled. Delay rate in winter is the highest, while Spring is lowest. In summer, the delay times
are highest in four seasons. The delay rate has no big difference in four seasons, while in
summer more flights are delayed by National Airspace System. And in fall, the delay flights is
much lower comparing to other three seasons.

Recommendations:

Based on our analysis, there are several suggestions we want to make to improve American
Airline On-Time Arrive performance to support Mission Objective pertaining to Customer
loyalty:
1. Among the high delay rate of carrier delay, carrier delay is due to aircraft cleaning, aircraft
damage, awaiting the arrival of connecting passengers or crew, baggage, bird strike, cargo
loading, etc. We can also have cooperation with other companies such as transfer customers to
corporate flight company when flight was delayed or canceled.
2. The top 3 airports with the highest delay frequency are DFW, ORD, MIA. Since American
Airline has a lot of delay flights at these three airports, we decide not to invest a lot of money in
these three airports. We would like to learn from Southwest, diverge our direction to some
nearby smaller airports, it can save us a lot of money on renting airport space. Also, it avoids
huge competitions between other airlines. American Airlines can save the money for more
customer loyalty program.
3. If American Airline is not willing to give up huge profits in those airports, then it should invest
heavily in getting access to scarce runway slots and terminal gates at some of the nations biggest
and busiest airports. Signing a long-term leasing agreement with the airport can give us priority
in getting space in the busy airport which can decrease customers gate to gate time to improve
our on-time performance.
4. As we discovered, a very high numbers of delayed flights caused by carrier delay were
manufactured by the same manufacturer: Mcdonnel-Douglas with model MD-82 and MD-83.
But since MD-82 also has a very good on time performance so we think American Airline should
not invest Mcdonnel-Douglas model MD-83 in the future. This type of aircraft seriously impacts
American Airlines on-time performance, which will cause American Airline keep losing its
long-term built reputation and customer loyalty. At the meantime, due to limit budget, American
Airline should not invest anymore new planes. They should do more maintenance on existing
planes like MD-82 to ensure its on time rate.

Based on our analysis, there are several suggestions we want to make to improve
American Airline On-Time arrival performance in order to support Mission Objective
pertaining to Customer loyalty:
As we discovered, a very high numbers of delayed flights caused by carrier delay were
manufactured by the same manufacturer: Mcdonnel-Douglas with model MD-82 and
MD-83. But since MD-82 also has a very good on time performance so we think
American Airline should not invest Mcdonnel-Douglas model MD-83 in the future. This
type of aircraft seriously impact American Airlines on-time performance, which will
cause American Airline keep losing its long-term built reputation and customer loyalty.
At the meantime, due to limit budget, American Airline should not invest anymore new
planes. They should do more maintenance on existing planes like MD-82 to ensure its
on time rate.
In the report we can see that the top 3 airports with the highest delay frequency is the
same in both origins and destinations. They are Dallas/Fort Worth International,
Airport O'Hare International Airport and Miami International Airport. which we think
are not surprising because these three airports are in the top 10 large hubs in United
States. They have a very large visitor flow rates which is a huge attraction to all airlines.
Also, DFW is always facing some extreme weather in Summer time which is causing a
lot of delay. But since American Airline is having so many delay flights in those three
airports, we think it is time to invest less on those airports. If we can learn from
Southwest, diverge our direction to some nearby smaller airports, it can save us a lot of
money on renting airport space, also it avoids huge competitions between other airlines.
American Airline can save the money for more customer loyalty program.
If on the other hand, American Airline is not willing to give up huge profits in those
airports, then it should invest heavily on getting access to scarce runway slots and

terminal gates at some of the nations biggest and busiest airports. signing a long
term leasing agreement with the airport can give us priority in getting space in
busy airport which can decrease customers gate to gate time in order to improve
our on-time performance.

Das könnte Ihnen auch gefallen