Defines a standard deviation classification method.
The StandardDeviation coclass represents dispersion about the mean, and this classification creates classes that represent this dispersion. The classes mainly have an interval that is either one whole or part (for example, a half or quarter) of a standard deviation. There will be one class (often labeled �the mean�) that will straddle the mean value by the class interval. Other classes will be adjacent to this on either side, representing increasingly disperse values from the mean. The classes will all have the same interval except for the lowest and highest classes that cover the endpoints of the data range.
As with other classification objects, you put values into the classification with IClassify::SetHistogramData. Then you must use the IDeviationInterval interface to specify the mean and standard deviation values. The TableHistogram object can be used here to calculate these values. Finally, you can produce the classes using the IClassify::Classify method.
The number of classes generated by IClassify::Classify is determined by the settings of the properties on IDeviationInterval, not by the value of the parameter passed to IClassify::Classify. However, you should still supply the parameter, as it will be modified to reflect the number of classes actually created.
This classification is intended for generally symmetric distributions of values that have a broad peak near the mean with the density of values diminishing away from the peak.
An example of a suitable map for this classification could be a population density or accident rates map. You would expect these values to have their greatest data density near a mean value, and values that vary significantly are scarce. The classic shape of this type of distribution is the bell curve.
When setting up a StandardDeviation classify object, you must set the Mean and StandardDev properties to be used for the class breaks before you call IClassify::Classify. By default, the classes will have an interval of one standard deviation. However, you can set the DeviationInterval property to give you more classes. The DeviationInterval property specifies what fraction of a standard deviation you want the class intervals to be. Typically, you would set this to be a half or quarter to give you twice or four times as many classes.
Windows, Solaris, Linux
|Provides access to members that control the classification methods.
|Provides access to members that control classification.
|Provides access to members that control the classification methods that require a standard deviation based range.