AI.Hack - a series of hackathons in Moscow and St. Petersburg

Hello, Habr! Very soon, hackathons dedicated to AI will be held in Moscow and St. Petersburg.

This is not the first time that artificial intelligence has become the subject of a hackathon, because the decisions that can be built with its help are essentially limited only by the imagination and skill of the teams.

This time the tasks are:

• Gazprom Neft
• Speech Technology Center
• Leaked
• BestFit.ai
• Robomed

But first things first.

Gazprom Neft

Since 2016, the company has been conducting detailed logging of customer transactions - payment by cards of gasoline and goods sold in small stores in the gas station.

Participants will have access to about 1.5 GB of data.

And here's what you need to do:

Prediction of customer churn

St. Petersburg

Using the data provided, create an algorithm that can identify among customers those who plan to stop refueling at Gazprom Neft gas stations. That is, it is necessary to create a predictive model of the user from horizons of 1 month. Of the options presented by the teams, 1 best will be selected.

Clustering customers based on the purchase of petroleum products and the search for patterns of consumption of related products and services

St. Petersburg

It is necessary to analyze and create the maximum possible number of Customer segments based on the purchase of petroleum products.

For each of the selected segments to find specific patterns of consumption of related products and services.

Dataset is the transactional sales flow of the Gazprom Neft gas station network that you use to solve the case.

• Sales NP
• Sales of STIU united in nomenclature groups

The task consists of two parts:

1. Find the maximum number of patterns in the purchase of petroleum products and divide customers into segments based on each pattern
2. Within the customer segments from paragraph 1, find the maximum number of patterns of size and the fact of purchase \ non-purchase of related products from the details to "EKTU Nomenclature Subgroup 4"

Task for Moscow

Moscow

Details will be known after March 12th.

Center for Speech Technology

Work is to be done with data obtained from surveillance cameras for pedestrian crossings

Intelligent visual inspection system

St. Petersburg

It is necessary to create a “smart surveillance camera”. It should monitor the movement of pedestrians and vehicles, record violations - speeding, red light, abnormal trajectories

Leaked

Rating system

Moscow

The system will be used to rank pharmaceutical manufacturers. The basis will be letters of seized drugs, taking into account the reasons for the seizure, recall and termination of treatment.

Information is not provided ready-made - instead, a list of open sources is given.

BestFit.ai

Classification of car images

Moscow

Development of a model for classifying car images according to various characteristics related to the peculiarities of human perception of car design. The task is at the intersection of machine learning and a new scientific direction - digital psychology. The solution will be developed on the basis of real data from our experimental database BestFit.ai

RoboMed

The company will present 2 tasks, one each for Moscow and St. Petersburg. Direction - analysis of the dataset of an anonymous sample of medical topics.

Where and when

St. Petersburg

March 9-11 (application deadline)
Gazprom Neft
Vilensky per., Building 14

Moscow

March 16-18 (collection of applications for participation - until March 14)
VEB Commune
2nd Donskoy proezd, 9

Register and find out more details here .