Pitney Bowes MapInfo Vertical Mapper - Manual (Page 168)

Spatial Correlation
166
Vertical Mapper 3.7
This Browser shows the results of a significance analysis where the specified weight
was 8.5.
Principal Components Analysis
Principal component analysis is an effective and popular method of multidimensional statistical
analysis. It provides a method both of reducing the number of variables to be analyzed by
eliminating redundancy and of detecting structure in the relationship between variables.
The basic principle of principal component analysis is that several correlated variables can be
combined into a single factor. In a very simple example, suppose there is a strong correlation
between the presence of a certain mineral in the soil and the presence of gold. The correlation can
be summarized in a scatter plot, and a regression line can then be fitted which represents the best
summary of the linear relationship between the two variables. If we can define a variable that
approximates the regression line, then that variable will capture most of the essence of the two
variables. The regression line can be used in future analyses instead of the two variables. Thus, two
variables have been reduced to a single factor.
Principal component analysis can be applied where there are multiple components. After the first
factor has been extracted, another can be found that defines the remaining variability, and this
process can be reiterated. However, as consecutive factors are extracted, they account for less and
less variability. It is therefore not particularly meaningful to continue past a certain point.
Performing Principal Components Analysis
1. From the Vertical Mapper menu, choose the Data Analysis > Spatial Correlation > Principal
Components command.
2. In the Principal Components Analysis dialog box, clear any open grids you do not want to include
in the correlation matrix.
3. In the Number of Components section, specify the maximum number of components you want
to find and the cumulative variance.
4. In the File Name box, type a new file name or accept the default.
5. Click the OK button.