relations between the two phenomena, depicted at different maps, dimensionless values
are used, called as correlation coefficient and correlation ratio. Correlation coefficient
r is calculated in the case when the relation between the phenomena А and В can be
considered straightforward; correlation ratio η serves to estimate curvilinear
dependence.
Numeric value r ranges from -1 to +1. If r is +1 or –1, it indicates, respectively,
the full direct or complete feedback. If r is close to 0, the relation between the
phenomena is absent. With r > = | 0,7 | the relation is considered essential. The value
of correlation ratio η varies from 0 to 1 and has the same properties. All elements of
the map content can be statistically studied. So that to obtain the most reliable results
of statistical survey of cartographical images, these studies should be carried out on the
maps of large scale, which accurately and in detail transmit all elements of the area,
but the opportunity of detecting some characteristics on the small-scale maps is not
excluded either.
Statistic research should be carried out according to the specific program. When
studying any element of a map, it is necessary to determine the size of the plot, the
method of sampling, the accuracy required for measuring. Determining the optimal
number of observations is an important condition for correct research. A small number
of observations can lead to gross results, and a rather big amount of observations is not
expedient from the economical point of view. The calculation of the correlation creates
the basis for formation of more complex types of analysis: regression, dispersion,
factor analysis, etc. Very often the study seeks to identify the main factors which
determine development, placement of a phenomenon. The task is solved by
bogodimensional factor analysis. It allows to minimize (to 2 or 4 main factors) a big
set of output indicators that characterize a complex phenomenon or a process [4]. The
use of mathematical modelling techniques in cartography is a progressive
phenomenon, but its value should not be overestimated. The use of techniques of
mathematical modelling require dismemberment of the objects into parts,
simplification of complex relations, removal of bi-effects, introduction of many pre-
requisites and restrictions etc. Formal mathematical methods which largely objectivize
the research results, should be combined with the qualitative analysis and objects
description with a clear geographic interpretation. If mathematic ways do not find
geographical interpretation, they are not taken into consideration as not reproducing
the position of objects, their peculiarities, ties etc.
The question of accuracy of map research belongs to most complicated and least
developed among other issues of using the mapping method.
It is relatively not difficult to determine the accuracy of measurements and
calculations by maps. This is done by cartometry and geometry of random errors. It is
much more difficult to state how accuracy is impacted by scientific-methodical
- 275 -