Network Graph Categorization Based on Features
Arts and Sciences
Arezu Mansuri, 0000-0003-3122-419X; Cody Packer 0000-0001-9385-6868; Prasuna Pillalamari, 0000-0002-5848-5261; Gabe Fragoso, 0000-0002-3150-2383; Frankie Vazquez 0000-0002-1292-699X
Having a large collection of varied network graph data is significant for research findings. We have revealed that complex networks of their respective categories (cheminformatic, ecology, and infrastructure network graphs) have distinct similar structural properties amongst themselves. The goal of this project is to be able to more effectively and accurately categorize different graph networks through various machine learning algorithms (logistic regression, lasso regression, linear SVC, decision tree, and random forest and obtained the most important feature of the graphs) based on underlying features within each respective category. In order to achieve a more accurate categorization, more graph features are being included in the machine learning algorithm. The tools we used are C++ for calculating features and python for parsing and organizing features.
Mansuri, Arezu; Packer, Cody; Fragoso, Gabe; Vazquez, Frankie; and Pillalamarri, Prasuna, "Network Graph Categorization Based on Features" (2019). Symposium on Undergraduate Research and Creative Expression (SOURCE). 834.