DBSCAN

DBSCAN (Density-Based Spatial Clustering of Applications with Noise) groups data based on the density of points. It identifies dense regions and expands them until no new points can be included within a specified radius. Points that do not belong to any dense region are classified as "noise" and are excluded from the clusters. This algorithm is excellent for detecting clusters with irregular shapes and ignoring outliers, but its effectiveness heavily depends on choosing the correct parameters.

EXAMPLE OF USE

uses
  UDBSCAN;
  
procedure ExampleClustering;
begin
  ShowArray(DBSCAN('C:\Delphai\Delphai\Datasets\Iris-Clustering.csv', 0.1, 5));
end;

CLASSES AND METHODS GUIDE

• function DBSCAN(aData : String; aEps: Double; aMinPts: Integer; aHasHeader : Boolean = True): TArray; or

• function DBSCAN(aData : TDataSet; aEps: Double; aMinPts: Integer): TArray; : defines the groups.

  • aData : CSV file path or TDataSet object that contains the data that will be used for training.

  • aEps : (Epsilon) determines the maximum radius around a point to consider others as neighbors, defining the necessary density to form clusters in DBSCAN.

  • aHasHeader : indicates whether the file has a header.