Module: ColumnsMatcher::Graph
- Includes:
- GRATR
- Defined in:
- lib/columns_matcher/graph.rb
Constant Summary collapse
- STUB_VALUE =
1000
Class Method Summary collapse
-
.build_dependency_graph(columns) ⇒ Object
column names as the keys and column values as the values Returns a graph of vertices and edges, with weights.
-
.get_mappings(pairings, first_graph, second_graph) ⇒ Object
pairings is a 2D array of sub-arrays of size 2 - the results of the Munkres assignment algorithm first_graph and second_graph are UndirectedGraph’s Returns mappings of columns with the keys as columns to match and values as columns to match against.
-
.match_graphs(graph_to_match_against, graph_to_match) ⇒ Object
graph_to_match_against and graph_to_match are are UndirectedGraph’s Returns mappings of columns with the keys as columns to match and values as columns to match against.
Class Method Details
.build_dependency_graph(columns) ⇒ Object
column names as the keys and column values as the values Returns a graph of vertices and edges, with weights
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 |
# File 'lib/columns_matcher/graph.rb', line 12 def self.build_dependency_graph(columns) combinations_of_column_names = Utilities::get_combinations(columns.keys) graph = UndirectedGraph.new combinations_of_column_names.map do |combination_of_column_names| first_column_name = combination_of_column_names.first second_column_name = combination_of_column_names.last first_column_values = columns[first_column_name] second_column_values = columns[second_column_name] if first_column_name == second_column_name entropy_of_column = Statistics::entropy(first_column_values) graph[first_column_name] = entropy_of_column else mutual_information_between_columns = Statistics::mutual_information(first_column_values, second_column_values) graph.add_edge!(first_column_name, second_column_name, mutual_information_between_columns) end end graph end |
.get_mappings(pairings, first_graph, second_graph) ⇒ Object
pairings is a 2D array of sub-arrays of size 2 - the results of the Munkres assignment algorithm first_graph and second_graph are UndirectedGraph’s Returns mappings of columns with the keys as columns to match and values as columns to match against
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 |
# File 'lib/columns_matcher/graph.rb', line 57 def self.get_mappings(pairings, first_graph, second_graph) first_graph_vertices = first_graph.vertices second_graph_vertices = second_graph.vertices mappings = Hash.new pairings.sort {|first, second| first.first <=> second.first }.each do |pair| first_graph_vertex = first_graph_vertices[pair.last] second_graph_vertex = second_graph_vertices[pair.first] mappings[second_graph_vertex] = first_graph_vertex end mappings end |
.match_graphs(graph_to_match_against, graph_to_match) ⇒ Object
graph_to_match_against and graph_to_match are are UndirectedGraph’s Returns mappings of columns with the keys as columns to match and values as columns to match against
40 41 42 43 44 45 46 47 48 49 50 51 |
# File 'lib/columns_matcher/graph.rb', line 40 def self.match_graphs(graph_to_match_against, graph_to_match) graph_to_match_against_vectors = Converter::convert_from_vertices_to_vectors(graph_to_match_against) graph_to_match_vectors = Converter::convert_from_vertices_to_vectors(graph_to_match) cost_matrix = Converter::convert_from_vectors_to_euclidian_distance_cost_matrix(graph_to_match_vectors, graph_to_match_against_vectors) pairings = Statistics::munkres_assignment_algorithm(cost_matrix) mappings = Graph::get_mappings(pairings, graph_to_match_against, graph_to_match) mappings end |