Primary classification

Previous Top

Means of primary classification are intended for estimating contents of the initial data for the purpose of selecting the optimal classifier and its customizing for execution of data import.

 

Primary classification includes:        

-formation of statistics on contents of input data;

-forming the list of unique values of a separate attributive characteristic;

-automatic adding the digital classifier.

 

Formation of statistics on the contents of input data and (or) automatic adding the digital classifier is carried out by means of a separate dialog window which is activated after pressing the button "Classification" of the main dialog.

Forming the list of unique values of a separate attributive characteristic is carried out by pressing the right mouse button over preliminary chosen column ("click" by the left mouse button) in the table of a set attributes.

For formation of statistics on contents of input data it is necessary to choose a file into which the generated protocol will be written down and to customize the protocol contents:

-List of sets (list of loaded names of sets);

-List of fields (list of unique names of attributive characteristics meeting in all input sets);

-Attributes of sets (formation of list of data sets with complete sets of attributes fields related to them will be executed);

-Numbering of lines (lines of the protocol will be numbered);

-Relations (after name of an attributive field after «=» the name of the semantic characteristic will be added, if such conformity preliminary has been set automatically by keys or manually);

-Only without ties (only the fields which do not have conformity with semantics) will be included into the list of attributes.

 

Automatic adding the digital classifier consists in automatic formation in the classifier specified on the basic bookmark of the main dialog, of the semantics corresponding to attributive fields for which the conformity is not assigned yet and in formation of objects for data sets for which conformity is not assigned yet.

Adding new semantics and objects is performed starting from the given code.

In case at adding another semantics (or object) it will be found  that there is already such a code in the classifier, search of the next "free" code will be executed.

When adding semantics to the classifier, the type and size of the attribute field are taken into account, the name of the field is written as the key and the name of semantics.

When adding an object to the classifier, the type of the data set (localization) is taken into account, the name of the set is recorded as the key and name.

 

Thus, for a primary estimation of contents of unknown data sets it is possible to create a new classifier (at a classifier choice to fill out a nonexistent name), to execute classification and to load the data. Then, by visual analysis of the loaded map, you can either determine the best of the existing classifiers and perform the conformity configuration, or assign the correct names to the automatically added characteristics (do not touch the semantics keys), and to the objects - the corresponding image.