Figure 15.
Solutions that, in the respondents’ opinion would contribute to the P&R parking use.
6.2. Binominal Logit Model
Logit models are used to explain qualitative variables depending on the level of exogenous
variables (qualitative or quantitative). They find important applications, among others in modeling
the finding of the unit of study in a certain state, determining the probability of occurrence of some
phenomenon. If the explained variable assumes two states, i.e., it indicates whether the studied
phenomenon occurred or not, we are dealing with a binomial model. In the analysis, the explained
variable took two values, hence the correct form was the binomial model. In addition, the binomial
logit model allows to find statistically significant factors determining the probability of occurrence of
some phenomenon and to examine the e
ffects of interactions between these factors, and an additional
advantage of the logit model is the ability to interpret its parameters. Accordingly, the data obtained
from the questionnaire was analyzed using a logit model. The logit model is used to determine the
probability of occurrence of a phenomenon. In this case, it is likely to use the P&R parking during travel.
If the explained variables are qualitative, their representatives in the model are zero-one variables
(dichotomous variables). These variables take values in the model [
83
]:
Y
=
(
1, i f the phenomenon occurs
(
the ob ject has the f eature
)
;
0, i f the phenomenon does not occur
(
the ob ject has no eature
)
.
(3)
where Y—dependent variable.
The logit model for the dependent variable Y is expressed as:
p
=
P
(
Y
=
1|X
1
,
. . . X
k
) =
exp
β
0
+
P
k
i=1
β
i
·X
i
1
+
exp
β
0
+
P
k
i=1
β
i
·X
i
(4)
where p—the probability of occurrence of a phenomenon, X
i
—independent variable (i
= 1 . . . k),
β
0
—free expression and
β
i
—logistic regression coe
fficient.
The logit model (4) is transformed using logarithm to the following form:
logit
(
p
) =
ln
p
1 − p
=
β
0
+
β
1
·X
1
+
β
2
·X
2
+
· · ·
+
β
k
·X
k
(5)
Equation (5) is called the odds ratio and is expressed as the ratio of the probability that a
phenomenon will occur to the probability that a phenomenon will not occur.
Energies 2020, 13, 3473
19 of 26
6.3. Logit Model Results
The questions included in the questionnaire allowed for the selection of many explanatory
(independent) variables, which were later used in the analysis to determine their impact (or no
impact) on the P&R parking use by respondents. At the beginning of the analysis, all characteristics
of the respondents were taken into account. To determine the probability of the P&R parking use,
the following variables were taken:
Y—P&R parking use,
X
1
—gender,
X
2
—age,
X
3
—education,
X
4
—monthly income (Gross) [PLN],
X
5
—the number of years of having a driving license,
X
6
—the average time spent traveling during a day,
X
7
—the average number of trips made during a day,
X
8
—the number of kilometers driven during a year,
X
9
—trip purpose.
Independent variables (X
1
, ..., X
9
) were selected using stepwise regression. Table
4
presents the
variables after selecting, which were taken for the assessment of the P&R parking in Cracow use (Y).
These variables characterized by a strong correlation with other variables, and a weak correlation with
each other. The authors developed three models to describe the factors determining P&R parking
use. These are models A, B, and C. The first model (model A) took into account only characteristics of
the respondents, i.e., gender, age, the number of years having a driving license, and monthly income.
The model’s form is presented below:
logit
(
p
) =
β
0
+
β
1
· X
1
+
β
2
· X
2
+
β
4
· X
4
+
β
5
· X
5
(6)
where the meaning of symbols was explained at the beginning of the subsection.
Do'stlaringiz bilan baham: |