db_clustering 0.1.13 → 0.1.14

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: fe813536c77416b8518d986832a137fbcc441aec
-  data.tar.gz: 5390301c390d4b026532f6d9bc7a5e97ebc7b405
+  metadata.gz: 615007829554b370c41083af36305a796ad5eb29
+  data.tar.gz: 9d0ae95d547512641bee5d8b4e59ab496ee15359
 SHA512:
-  metadata.gz: ff53b70b7505433199f3b6b4e5df47c48e39af1d915d14db1489c56ce12112dd8b01522dfb5b0aab892eb2f6379c95141ddfa40aba97c0c11037a63a2e81ad54
-  data.tar.gz: abaac0ab99a4edfa2f0caca7a1de352ae2628168aa870e4caee59a31dc1c690db5e6a2efc19f6fda6046ced0613eebcadabcb858624ad9d29dd5bf10a523e533
+  metadata.gz: 97545dae398e584b13884969e9219212301be23e578b0cb8f0a210f70445fae9e5e04d8e150c8de75fa02d9838e3881ed102e7a6cc3caaee7326f60b01b98d72
+  data.tar.gz: 8021b6cb3c55745a851ccb5c011ebd79717dc232b51e4e6c1d9ddf2b1b5e4eaf25b5e75629700181ee5bed2fff577852f5428748054a0221a76383d332f7889c

data/VERSION

@@ -1 +1 @@
-0.1.13
+0.1.14

data/lib/algorithms/density_based/dbscan.rb

@@ -6,16 +6,21 @@ module DbClustering
 
       attr_accessor :datasource, :clusters
 
-      def initialize(datasource:, distance_metric:)
+      def initialize(datasource:, distance_metric:, debug: false)
         @datasource = datasource
         @distance_metric = distance_metric
         @clusters = []
+        @debug = debug
       end
 
-      def cluster(max_distance:, min_neighbors:, debug: false)
+      def cluster(max_distance:, min_neighbors:)
         @clusters = []
         cluster = nil
 
+        if @debug
+          last_printed_progress = 0.0
+        end
+
         @datasource.iterate_all_points do |point, current_index, points_count|
           neighbors = @datasource.neighbors(point: point, distance_metric: @distance_metric, max_distance: max_distance)
 
@@ -24,22 +29,43 @@ module DbClustering
           elsif point.cluster.nil?
             cluster = DbClustering::Models::Cluster.new
             @clusters << cluster
+            cluster.add(point)
+
+            expand_cluster(point, neighbors, max_distance)
+          end
 
-            neighbors.each do |neighbor|
-              if neighbor.cluster.nil?
-                cluster.add(neighbor)
+          yield(point, current_index, points_count) if block_given?
+
+          if @debug
+            point_type_string = point.is_edge_point? ? 'E' : point.is_core_point? ? 'C' : 'N'
+
+            if point.cluster
+              print "(#{point_type_string}|#{point.cluster.id}|#{point.cluster.points.count})"
+            else
+              print "(#{point_type_string}|nil|0)"
+            end
 
-                if debug
-                  print "+"
-                end
+            progress = (current_index + 1) * 100 / points_count.to_f
 
-                # add the neighbors of the neighbor to the neighbors to fully expand the cluster
-                neighbors |= @datasource.neighbors(point: neighbor, distance_metric: @distance_metric, max_distance: max_distance)
-              end
+            if progress > last_printed_progress + 1
+              print "[#{progress.to_i}%]"
+              last_printed_progress = progress
             end
           end
+        end
+      end
 
-          yield(point, current_index, points_count) if block_given?
+      def expand_cluster(point, neighbors, max_distance)
+        neighbors.each do |neighbor|
+          if neighbor.cluster.nil?
+            point.cluster.add(neighbor)
+
+            if @debug
+              print "+"
+            end
+
+            neighbors |= @datasource.neighbors(point: point, distance_metric: @distance_metric, max_distance: max_distance)
+          end
         end
       end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: db_clustering
 version: !ruby/object:Gem::Version
-  version: 0.1.13
+  version: 0.1.14
 platform: ruby
 authors:
 - Cihat Gündüz