Skip to content

Graph linking strategies - multivariate

LinkGraphs

Builder class for linking multiple graphs, through which we can access linking strategies.

Attributes:

  • graph: networkx.Graph object
  • graphs: dictionary of networkx.Graph objects
  • command_array: an array that stores linking strategies

correlation(corr)

Notes that we want to connect graphs in a multivariate graph based on correlation.

correlation_sliding_window(corr)

Notes that we want to connect graphs in a multivariate graph based on correlation for sliding window graph.

dynamic_timewarping()

Notes that we want to connect graphs in a multivariate graph based on dynamic time warping.

positional_correlation_sliding_window(corr)

Notes that we want to connect graphs in a multivariate graph based on correlation for same positioned windows of series.

sliding_window()

Notes that we want to connect graphs in a multivariate graph to create sliding window graph.

time_cooccurrence()

Notes that we want to connect graphs in a multivariate graph based on time co-ocurrance.

StrategyLinkingMultipleGraphs

Links multiple graphs together.

Attributes:

  • graph: networkx.Graph object
  • graphs: dictionary of networkx.Graph objects
  • strategy_precedence: tells in which order should the strategies be excetuted

StrategyLinkingMultipleGraphsByTimeCooccurrence

Bases: StrategyLinkingMultipleGraphs

Links nodes from multiple graphs based on their sequential order.

StrategyLinkingMultipleGraphsDynamicTimeWarping

Bases: StrategyLinkingMultipleGraphs

Connects visibility graphs based on Dynamic time warping.

StrategyLinkingMultipleGraphsSlidingWindow

Bases: StrategyLinkingMultipleGraphs

Sequentially links graphs made by sliding window mechanism.