SAS Sample Pre

Database Management and Modeling
Final Project
CHENYE PAN, EN-CHIAO WEN
JIALIN ZHAO, WENQIAN WANG, XUETING SHEN
DECEMBER 7, 2015

AGENDA
Background
ObjecRves
Methodology
Findings
Conclusion

SFO conducts a yearly comprehensive survey of
their guests to gauge saRsfacRon with their
faciliRes, services, and ameniRes.

Customer interviews held at all airport terminals
and boarding areas from May 11 through May 26,
2011
Data Source

•  Understand and monitor passenger purchase
behavior and percepRon of SFO
•  IdenRfy factors that have strong impact on in-terminal
purchase and SFO saRsfacRon
•  Understand what diﬀerent passenger groups expect
of SFO
•  IdenRfy areas to improve
ObjecRves

This data set contains 3188 observaRons and 7 core
variables, including two dependent variables:

•  Purchase in in-terminal retail stores
•  Overall saRsfacRon

Other core variables are “Age”, “Gender”,
“Household Income”, “Times ﬂown out from SFO”
and “Trip Purpose”.
Variables

Purchase

Cross-TabulaRon Analysis
Logit Regression
Sa+sfac+on

Index
Cross-TabulaRon Analysis
Logit Regression
Approach
This report’s main purpose is to analyze which passengers are
more likely to purchase in SFO and which have higher saRsfacRon
with SFO. StaRsRc measures and model are used.

Purchase
Age, Flying Times and Trip Purpose have
no signiﬁcant associaRon with Purchase

•  The Chi-Square staRsRcs is signiﬁcant (p<0.05)
indicaRng an associaRon between Gender and
Purchase in retail stores such that the female
passengers are more likely to purchase in in-
terminal retail stores (40% versus 33%)
Measure of AssociaRon
Gender and Purchase
•  Gender 0=male, gender 1=female
Purchase 0=no, purchase 1 =yes

•  Odds raRo ≅1.3, indicaRng that females are 1.3
Rmes more likely than males to purchase
Gender and Purchase
•  Gender 0=male, gender 1=female

Group 1
Group 2

indicaRng an associaRon between Household
Income and Purchase in retail stores such that
the passengers with over 100K household income
are more likely to purchase in in-terminal retail
stores (40% versus 35%)
HHI and Purchase

•  HHI 0 =$100K and under, HHI 1=above $100K
Purchase 0 =no, purchase 1 =yes

•  Odds raRo (OR) ≅1.25 means passengers with
HHI >100K are 1.25 Rmes more likely than
passengers with HHI ≤$100K to purchase.
HHI and Purchase

•  HHI 0=$100K and under, HHI 1=fabove $100K
Group 1
Group 2

Gender Income
Logit
(Log of odds)
Odds of non-
purchase
Male (0) HHI ≤100K (0) 0.8028 e0.8028=
2.2318
Male (0) HHI >100K (1) 0.8028+(-0.2618)=0.5410 e0.5410
=1.7177
Female (1) HHI ≤100K (0) 0.8028+(-0.3101)=0.4927 e0.4927
=1.6367
Female (1) HHI >100K (1) 0.8028+(-0.3101)+(-0.2618)=0.2309 e
0.2309
=1.2597
!
Logit (purchase) = 0.8028 - 0.3101*Gender (Female) -
0.2618*HHI (>$100K)
•  Female passengers whose HHI is
more than $100K are most likely to
purchase in the SFO retail stores
•  Male passengers whose HHI ≤
$100K is least likely to purchase in
SFO retail stores
Logit Regression Model

SaRsfacRon
Age, Gender and Household Income have
no signiﬁcant associaRon with Purchase

indicaRng an associaRon between Flying
Frequency and Overall SaRsfacRon such that the
passengers who have ﬂown more than 6 Rmes
from SFO in the past 12 months are more likely to
have low saRsfacRon with SFO than those ﬂown
1- 6 Rmes (31% versus 24%)

Flying Times and SaRsfacRon
•  Times 0 =1-6 Rmes, Rmes 1 =more than 6 Rmes
saRsfacRon 0 =low, saRsfacRon 1 =high

•  Odds RaRo (OR) ≅0.7, indicaRng that Group 1
(those ﬂying 6 Rmes and less) are 30% less likely
to have low saRsfacRon than Group 2 (those
ﬂying more than 6 Rmes)

Group 1
Group 2

Flying Times and SaRsfacRon
•  Times 0 =1-6 Rmes, Rmes 1 =more than 6 Rmes

indicaRng an associaRon between Trip Purpose and
Overall SaRsfacRon such that the passengers who
ﬂy for business are more likely to have low
saRsfacRon with SFO than those who ﬂy for non-
business (30% versus 22%)

Trip Purpose and SaRsfacRon
•  purpose 0 =business, purpose 1 =non-business

•  Odds RaRo (OR) ≅1.5, indicaRng that Group 1
(those ﬂying for business) are 1.5 Rmes more
likely to have low saRsfacRon than Group 2
(those ﬂying for non-business)

Group 1
Group 2

Trip Purpose and SaRsfacRon
•  purpose 0 =business, purpose 1 =non-business

Logit Regression Model
Purpose Times
Logit
(Log of odds)
odds of Low
Satisfaction
Business (0) ≤6 times (0) -0.8940 e-0.8940
=0.4090
Business (0) >6 times (1) -0.8940+ 0.2642=-0.6298 e-0.6298
=1.8772
Non-business (1) ≤6 times (0) -0.8940+(-0.3939)=-1.2879 e
-1.2879
=0.2758
Non-business (1) >6 times (1) -0.8940+0.2642+(-0.3939)=-1.0237 e
-1.0237
=0.3592
!

•  Passengers who travels for
business and have flown out
from SFO for more than 6 Rmes
are least saRsfied with SFO.
•  Passenger who travels for non-
business and have flown out
from SFO for 1- 6 Rmes are
most saRsfied with SFO.
Logit (low saRsfacRon) = -0.8940 - 0.3939*Purpose (non-
business) - 0.2618*Times (>6)

•  Female passengers and high-income passengers are more likely
to purchase in SFO
Ø  SFO may provide more products and brands that cater to
high-income females

•  Business passengers and passengers who have flown more than
6 Rmes out from SFO in the past 12 months are less saRsfied
with SFO
Ø  Further research is needed to find out why business
passengers have low saRsfacRon and find a way to
improve the returning passengers’ saRsfacRon with SFO
Insights

Areas to Improve
•  Artwork and exhibiRons
•  Restaurants
•  Retail shops and concessions
•  InformaRon booths (lower level near baggage claim)
•  InformaRon booths (upper level - departure area)
•  Signs and direcRons on SFO airport roadways
•  Airport parking faciliRes
•  Long term parking lot shutle (bus ride)

THANK YOU!
&
GOOD LUCK WITH YOUR FINALS

SAS Sample Pre

More Related Content

Similar to SAS Sample Pre (7)

SAS Sample Pre