Read Results from Neural Network Module

Select 'Read Results from Neural Network Module...' from the ArcSDM3 menu.

Select the unique conditions grid theme with which you want to associate the results from the neural network analysis. This is the same unique conditions grid theme that was used to generate the neural network input files.

Click OK.

You will be prompted to do the following:

  1. select the RBFLN Result File (.rbn) to read. If the RBFLN analysis was not done, click 'Cancel' to bypass this dialog.
  2. select the Fuzzy Classify Result file (*.fuz) to read. If a fuzzy classification was not performed, click 'Cancel' to bypass this dialog.
  3. select the PNN Results file (.pnn) to read. If the PNN analysis was not done, click "Cancel" to bypass this dialog.
  4. specify the name and location to create a dBase file to write the results from the neural network analysis. The default name is SDMUC#_nnrslt.dbf.
  5. the user then needs to symbolize of the desired result field of the response theme. The possible fields are those of the neural network result file (SDMUC#_nnrslt.dbf). Symbolization of the results from GeoXplore have been a problem. It has been found that symbolization works best using the quantile method with a larger number of classes. Fifteen to 20 classes is often a good starting place.Descriptions of each are provided in the description of the table that follows.

The Neural Network Result file
Field Name Field Alias Data Type Size (Precision) Description
ID ID Long Integer 6 Unique condition identifier. Used to join the results to the unique conditions attribute table.
RBFLNMmshp Pattern Membership Decimal 10 (6) The result from the Radial Basis Function Link Network (RBFLN). Values fall in the range 0 to 1. 1 indicates a mineral occurrence and 0 indicates the absence of a mineral occurrence. The higher the value the more closely the unique condition resembles a pattern in which 'deposit' training sites are known to occur. Values outside the range 0 to 1 result when the classification SSE is poor in DataXplore.
FuzzyClstr Fuzzy Cluster     The result from the Fuzzy Classification. The class to which the unique condition was assigned.
Pat# Pattern <#>     A series of fields, one for each class defined during the fuzzy clustering. The value is a measure in the range 0 to 1 of the fuzzy membership of closeness of the unique condition to that particular pattern or class.
PNN PNN Decimal   The results of the Probabilistic Neural Network (PNN). (RBFLN). Values fall in the range 0 to 1. 1 indicates a mineral occurrence and 0 indicates the absence of a mineral occurrence. The higher the value the more closely the unique condition resembles a pattern in which 'deposit' training sites are known to occur. Values outside the range 0 to 1 result when the classification SSE is poor in DataXplore.

 

Next Section Contents Home