A key concept for understanding the tests used in logistic
multiple regression is that of a log likelihood. Usually,
though, overall significance is tested using Model Chisquare,
which is derived from the likelihood of observing
the actual data under the assumption that the model that
has been fitted is accurate. Table 2 contains the base model
results for the logistic regression analysis.
Using the multivariate logistic regression method, the
spatial relationship between landslide-occurrence location
and landslide-related factors was calculated. For the implementation
of the multivariate logistic regression method,
SPSS program was used for the calculation of the correlation
of landslide to each factor. First, all of the 10 factors that
were constructed in the spatial database were used. Then,
logistic multiple regression coefficients of the factors were
calculated (Table 2). The coefficients of the logistic multiple
regression model is estimated using the maximum-likelihood
method. In other words, the coefficients that make the
observed results most “likely” are selected. Since the relationship
between the independent variables and the probability
is nonlinear in the logistic multiple regression model,
an iterative algorithm was used for parametric estimation
(Dai and Lee, 2002). In Table 2, two types of correlation
can be seen; one is positive coefficients such as slope and
the second is negative coefficients such as curvature. After
calculating the logistic regression coefficients of the 10 landsliding
factors, formulae (3)–(5), were used for the landslide
hazard mapping for three test areas as shown below.