Information Technology Reference
In-Depth Information
3
Validation of the Payment Simulation
3.1 Calibration
In order to run and validate the payment simulation, we have to calibrate the
input parameters with initial values. These values based mainly on German
studies about the payment behavoir and real data from available online stores.
First, our model contains the four payment methods: invoice (
Z 1 ), prepay-
ment (
Z 4 ). These types of payment
are widely-used in Germany. Furthermore, we set the number of customers to
100.000. In order to study the gender-specific payment behavior, we use the
following gender ratio female/male: 23/77 and 77/23. Next, we define the func-
tion
Z 2 ), credit card (
Z 3 ),andcashondelivery(
by the co-domain shown in Table 1. The values are adopted from the
study [2]. Based on these gender-unspecific values, we calculate the function val-
ues for men (
fav
fav m )andwomen(
fav f ). For calculating these values, we need
the ratio
of the favored payment methods in dependency of the gender. The
ratio is shown in Table 2 as the result of an analysis of real shop data. Further,
we need the
r
function as an input parameter. The values of the distinct
payment types are defined in Table 3. These values are derived from a study [3].
In this study customers evaluated different payment methods with the grade 1
(very well) till 5 (bad). The utility value is the percentage rate of the customers
who gave a grade better than average grade 2.6.
utility
3.2 Simulation Run and Validation
We run the simulation experiment 30 times with the input parameters described
in the section above, ten with a gender ratio of 23/77, ten with 50/50, and ten
with 77/23. Due to the lack of space, we only show the simulation results pro-
duced with a gender ratio 23% female and 77% male. In our evaluation (see
Sec. 3.3), we also include the inverse and fifty-fifty gender ratio and. The av-
erage of the simulation results is shown in Table 4. We validate the results by
using the values from this study [2] and real store data. The study includes seven
configurations of payment methods that are shown in Table 5. If we compare
the results of the study with the results of the proposed simulation model, then
four configurations exactly coincide and only three configurations have differ-
ences. A detailed comparison shows in Table 6 the absolute and relative errors.
Table 1. Probability distribution of the favorite payment method depending on the
gender
Gender
Invoice ( Z 1 )Prepayment
( Z 2 )
Credit Card
( Z 3 )
Cash on
Delivery ( Z 4 )
Cancel
neuter
0.65
0.04
0.23
0.03
0.05
male
0.62
0.04
0.26
0.03
0.05
female
0.77
0.03
0.13
0.02
0.05
Search WWH ::




Custom Search