How was the first academic year Data Mining Track

On October 5, Data Mining Labs Data Mining Lab launches a new semester. What it is, how to get there and what opportunities students get - welcome under cat.

The idea of ​​launching our program arose from the desire to create a full-fledged course that would allow, on the one hand, to understand how and why data mining methods are used, and on the other, how to put them into practice. In February 2013, we launched a pilot version of such a project, called the Data Mining track, based on the GameChangers program (St. Petersburg). To participate in the program, we were able to attract experts from large companies (IBM, EMC, Siemens, Yandex, etc.) and representatives of universities in St. Petersburg.

As a result, we had 19 experts, including those from the United States (Microsoft, Denver) and Germany (fortiss Institute of Robotics at the Technical University of Munich).

The course was based on lessons from experts from various industries (lectures, round tables and homework), intensive teamwork, individual quests and projects.
')


Representatives of the companies talked about their experience in data analysis and building solutions based on Big Data. For example, experts from Siemens conducted a class on signal analysis for maintaining the security of dams and electrical networks, Yandex for search technologies, Yota Bank for scoring algorithms, HP Labs for pattern recognition, etc.



As quests, the guys including preparing articles for Habr on the tasks they were doing as part of homework: preprocessing scoring data and solving the problem of recognizing handwritten images using Python and scikit .

At the stage of teamwork, we divided the children into 2 groups and gave each topic of the project and TZ for a period of one month. We simulated the situation of working with a capricious customer, who set goals very crookedly and periodically did not get in touch. Experienced techlides from Netrika and Yandex helped students cope with this stressful task.

As a result, students completed 2 team projects:
1. Analysis of the company in order to identify key persons who are able to “lead” a team or group of people. The project was based on the analysis of social user profiles. network My Circle.
2. Development of a recognition system for real accounts of specific people on Twitter.



Doing graduation work was the last part of the training. At first, the students presented the topics of their research to the experts who gave feedback on the adequacy and relevance of the task. After that, some experts wished to become mentors and supervise the preparation of the final work of our student. Among the interesting research topics, the project of predicting patterns of behavior of the VK user (we will write a separate article about this project) and the creation of a model for predicting bankruptcy of enterprises deserve special attention.



In many ways, this year was experimental (we mocked the students and they did everything they could for us)), but it helped us to revise the entire training system as a whole: now we have allocated separate courses in Big Data, Machine Learning, R; we want to test gamification (about this we can also prepare a separate article); revised role model in the classroom, etc.

Now we are actively recruiting for a new academic year as part of the already independent Data Mining Labs program.
If the topic of our educational program is interesting to the community, then we will be happy to continue to reveal the details of the past year of study and share news about the current course.
As authors of any undertaking, we want to hear comments on the program and advice from more experienced colleagues. Wishes and constructive criticism are accepted and welcomed!