akashjborah97 / graph-based-feature-selection-for-dimensionality-reduction Goto Github PK

- Graph Based Feature Selection is a new approach of reducing the dimensionality of a dataset using a Graph Based approach. - The apporach tries to generate a Kruskal's minimum spanning tree of a graph where the features of the dataset are the vertices and the correlation among them are the weights of the edges. -The edges having weights greater than the user defined threshold are removed. Hence, reducing the dimension of the dataset.