Class: SameSame::DbscanNeighborhood

Inherits:
Object
  • Object
show all
Defined in:
lib/same_same/dbscan_neighborhood.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(attrs = {}) ⇒ DbscanNeighborhood

Initializes algorithm with all data that it needs.

* points - points to cluster
* eps - distance threshold value
* minPoints - number of neighbors for point to be considered a core point.



21
22
23
24
25
26
27
28
29
30
31
32
33
 
# File 'lib/same_same/dbscan_neighborhood.rb', line 21

def initialize(attrs = {})
  self.eps    = attrs.fetch(:eps)
  self.points = attrs.fetch(:points)

  build_index_mapping
  
  vector_calculator    = attrs[:vector_calculator] || DbscanNumericVectors.new
  distance             = attrs.fetch( :distance )
  use_term_frequencies = attrs[:use_term_frequencies] || false

  self.adjacency_matrix = 
        calculate_adjacency_matrix(distance, points, vector_calculator)
end

Instance Attribute Details

#adjacency_matrix ⇒ Object

Contains distances between points.



4
5
6
 
# File 'lib/same_same/dbscan_neighborhood.rb', line 4

def adjacency_matrix
  @adjacency_matrix
end

#eps ⇒ Object

Threshold value. Determines which points will be considered as neighbors. Two points are neighbors if the distance between them does not exceed threshold value.



9
10
11
 
# File 'lib/same_same/dbscan_neighborhood.rb', line 9

def eps
  @eps
end

#index_mapping ⇒ Object

used to cache index of points in the matrix



12
13
14
 
# File 'lib/same_same/dbscan_neighborhood.rb', line 12

def index_mapping
  @index_mapping
end

#points ⇒ Object

Returns the value of attribute points.



14
15
16
 
# File 'lib/same_same/dbscan_neighborhood.rb', line 14

def points
  @points
end

Instance Method Details

#neighbors_of(p) ⇒ Object 



35
36
37
38
39
40
41
42
 
# File 'lib/same_same/dbscan_neighborhood.rb', line 35

def neighbors_of( p )
  Set.new.tap do |neighbors|
    i = index_mapping[p]
    (0..index_mapping.size-1).each do |j|
      neighbors.add(points[j]) if adjacency_matrix.lookup(i,j) <= eps
    end
  end
end