Machine learning in Avito. Video recordings from the Data Science Case Club meeting on March 14

Last week, the Data Science Case Club meeting took place, at which Avito experts talked about which business tasks of the service are solved using machine learning. In particular, we talked about recommendations, contextual advertising and moderation. Under the cut more details about the meeting and video reports.


The Case Science Data Club in Avito is the third meeting organized by the HSE FKN teacher Peter Romov for students enrolled in the “Machine Learning in Industry” elective. We asked him to tell more about how the idea to hold such meetings appeared and how they differ from other events from the world of Data Science.

“Data Science Case Club is an attempt to bridge the gap between industry and university courses.
The FKN specialization in machine learning , for example, includes such disciplines as applied statistics, machine learning, in-depth training, big data processing. Each course understands examples from real life. But there is no such course that would collect all the knowledge in a single chain and show how to create business value with their help.
The task of the case club is to show on the example of companies from different industries. How can data optimize business performance? How does a concrete solution go through all the stages from problem statement to implementation?
At the case club, we will not analyze in detail the approaches to solving the problems already posed of machine learning. Instead, we will focus on the problems associated with the formulation of the problem, the assessment of the economic effect, the introduction of models into production and their support. ”

The meeting was opened by Ivan Goose, head of analytical research at Avito, who gave an overview of the areas in which analytics and machine learning are used in Avito, and also revealed some details about the internal analytical processes in the company. Further, the individual cases were reviewed in detail. Andrey Rybintsev, Andrey Ostapets and Vasily Leksin told (respectively) about moderation and quality of content, Avito.Context and service recommendations. Video recording of the meeting below, slides of the presentations can be found on the link .
')


If you want to learn more about the technical features of the implementation of contextual advertising and service recommendations in Avito, then we have already told it at last year's Python Data Science mitap. Videos are available on our channel.

• “ Contextual advertising in Avito: what's under the hood? »Vadim Ayuev and Andrey Ostapets
• “ Building a Recommended System in Python ” Vasily Leksin

If you have any questions - ask them in the comments. Video recordings of previous meetings of the case club can be found here .