Coverage algorithms for objects of urban environment
On March 11 was held another seminar and workshop within the framework of our Research Group "Machine Learning and Social Computing". The first part of the event was devoted to the second part of the workshop of topic modeling and text mining. Next, Irina Krylova and Stanislav Kozlov told about visualization geocodes in R and Python. The main participants of the event were participants of our Research Group, students, undergraduates, post-graduate students and teachers of BAs "Sociology and Social Informatics" and "Political Science", as well as students of Minor "Data analysis".
With the use of data from Russian news website, it was possible to get more familiar with LDA algorithm (Latent Dirichlet allocation). The results allowed to correlate the resulting topics with the original categories of publications on the news website, and also to track the dynamics of the popularity of topics over time. It was really fascinating for listeners to work with the popular and relevant news, that's why almost all participants were engaged in the discussion of work.
In the second part of the seminar, Irina Krylova (participant of our Research Group), told about the basic methods of working with geodata and demonstrated various possibilities of visualizing objects on the map in R. Ira showed the techniques of working with data that allow not only to trace the geographical coordinates of houses, restaurants, hotels or other establishments but also map various characteristics related to their functions.
Further, Stanislav Kozlov (participant of Our Research Group) presented the results of his work on optimizing the distribution of transport routes and uniting destinations in the Express Delivery company. He told about how with the use of Python it is possible to take into account the distortions of the geographic map for object clustering. Stanislav also presented a network of geographic facilities relevant to his profession.