Using library model
Not Lite
Classify groups using library model routine classifies groups of points using a classification model prepared and distributed by Terrasolid. Library models are distributed with the setup package, and extracted to \machine_learning application subdirectory in the installation. A single library model consists of a TXT file and a matching PKL file.
A library model has default target classes for each result object. You can modify the target class mapping after selecting the desired result row.
SETTING |
EFFECT |
|---|---|
Groups with |
Determines which groups are classified: •One or more points - groups with one or more points in the source class. •Majority of points - groups with the majority of points in the source class. •All points - groups with all points in the source class. |
From class |
Source class(es) from which to classify points. |
|
Opens the Select classes dialog which contains the list of active classes in TerraScan. You can select multiple source classes from the list that are then used in the From class field. |
Confidence |
The target attribute field for storing the estimated confidence of the classification. •Do not store - discard the probability values after the classification. •Reliability - store the group's classification confidence to the TerraScan reliability attribute. •<Extra attribute> - store to suitable extra attribute. |
Inside fence |
Determines how a fence or selected polygon(s) effect the classification: •Not used - fence or selected polygons are ignored. •One or more points - groups are classified if one or more points are inside. •Average xy - groups are classified if the average xy point is inside. •Majority of points - groups are classified if the majority of points is inside. •All points - groups are classified if all points are inside. |
Model |
Text file that stores the model details. Use the Browse button in order to open a standard dialog for opening a file. |
Edit |
Open Class mapping dialog for modifying the classification target class for the selected result row. |
