SIBUR Challenge 2019 – industrial data analysis competition

Hi all!

The online stage of the data analysis competition SIBUR Challenge 2019 continues.

Briefly about the main thing:

• The SIBUR Challenge is our signature hackathon, which we do together with the AI ​​Community. As cases, we use real production tasks based on real data.
• The prize fund is 1 rubles, plus vacancies and internships for the winners.
• You can join the race until November 17, the offline stage will be held on November 23-24 in Moscow.
• So far, over 1200 participants have registered.

Tasks are divided into two groups:

• The first is about business: it is necessary to predict the market value of products that are important for the industry;
• The second is about production: it is necessary to predict the activity of the catalyst that participates in the polymerization process (you can read about other processes in petrochemistry in article Alexey Vinnichenko in our blog).

The rest is under the cut.

About stages

The competition takes place in two stages:

First stage - online: October 21 - November 17

At this stage, the participants will have to develop the best solutions to the problems (break through the baselines). And also there is an opportunity to attend webinars, earn points, communicate and get to know the participants. Individual participants can join teams of up to 6 people before or after registration. By the way, after registration, a private chat in telegram will become available, where you can chat, find a team and ask clarifying questions. The line-up of the team may change at the online stage, but closer to the finals, it is necessary to decide on the line-up - then it will be impossible to change it.

Second stage – offline: November 23 — 24, Moscow

At this stage, it is necessary to develop a prototype of a digital product, which is based on the models created at the previous stage, to discuss issues of interest with mentors and organizers. It will be possible to communicate with experts from SIBUR, who have been making cool decisions based on advanced analytics for 2 years, as well as with the HR team. Here, the finalists will be able to exchange earned points for prizes: quadcopters and sweatshirts. There will be an informal party and buffet!

Who is it all for?

You will be interested in the Sibur Challenge if you:

• data engineer,
• data analyst,
• developer.

There are two streams in the competition:

• student - only students of Russian universities can participate,
• the main one is students and specialists.

To reach the final, you need to enter the top 10 in one of the streams.

Finally, about tasks

All tasks are divided into two groups: market modeling and production optimization. A prerequisite for the stage is the solution of two main data science problems.

First group:

Prediction of market prices for important products and semi-finished products in the industry - polyethylene terephthalate (PET, main task) and synthetic rubber (additional task). A qualitative forecast of these prices will allow SIBUR to adjust the production of rubber and PET to market conditions and maximize profits from their sale.

The second group:

Optimization of the operation of a cascade propylene polymerization unit. In the main task, it will be necessary to learn how to predict the future activity of the catalyst used in the polymerization process, and in the additional task, to predict the production of by-products. These models will help SIBUR optimize catalyst utilization and increase polypropylene production.

A few more comments...

Bot and private chat in telegram:

Registration in telegram gives access to information about accrued points. The bot notifies about the stages of the competition, and in the chat there is an opportunity to find the missing participants in the team, if necessary, ask questions to the organizers.

Teams:

The creation of a team is not a prerequisite for the online stage, but a requirement for reaching the final. We noticed that one participant rarely possesses the entire range of skills at a sufficient level necessary to solve the tasks of the final stage.

Gamification:

You can earn points by actively participating in the competition - for improving the solution, solving additional tracks, attending webinars and much more. You can spend these points on additional data for tasks, on consultations with the SIBUR team, as well as on gifts - branded merch and gadgets. The first 500 bonus points will be given to those who have registered the earliest.

A little video:

Welcome word Alexey Vinnichenko, leader of Advanced Analytics at SIBUR.

You can register by link.

Source: habr.com