Analysis of the Scientific and Pedagogical NRU HSE Staff Publication Graph

This work was performed to study the occurrence of articles of lecturer in collaboration with a student depending on the structural information of the co-author publication graph and such attributes as a duration of scientific cooperation, status and position of a lecturer, as well as expose the grades for the course project. The following problems were considered: analysis and processing of data for scientific cooperation with students within the term papers and graduation works of qualified employees and students of the NRU Higher School of Economics. We build a graph with vertices as the names of students and teachers. Two vertices are connected by an edge, if a student wrote the course project under the guidance of a teacher. Reinforcement counts a personal information about the lecturers and attachments of scientific articles written by the teacher in collaboration with the student. In the framework of the conducted work we solve the problem of binary classification using logistic regression for continuation of collaboration for the next year. According to the results of the logistic regression ROC curve is constructed estimating the prediction quality.