YRP MSU & Microsoft
MSU BigDATA ImagineHack
Take part in Hackathon MSU
From 5 to 6 March Lomonosov MSU will host MSU BigDATA Imagine Hack Hackathon, which is dedicated to the construction of predictive models based on big data and is dedicated to the international student competition of IT projects imagine Cup.

MSU BigDATA Imagine Hack is conducted by the youth research platform of MSU with the support of Microsoft. Students, Masters and graduate students of Moscow University can take part in the participation

The winner will receive the prize and pass to the regional final Imagine Cup. For the best projects, the Youth research platform of MSU will allocate its own direction for further work on the project in the Alma mater walls. Each team will receive $100 to use Azure services

Conditions of participation:
1. Imagine Cup-Team competition, get together a team of up to 3 people.
2. The ability to use Microsoft Azure Machine Learning, Microsoft Cognitive Services, or Microsoft PowerBi in the development of technologies.

On the Hackathon you will be able to use one of the following data sets:
Chemometrics to determine the taste of wine and coffee
Chemometrics was born as a chemical discipline using methods of statistics, applied Mathematics and informatics for extracting useful information from the measured chemical data and allowing to optimize analytical chemical Processes.

Chemometrics allows to give answers to non-trivial questions about complex systems with a lot of factors.

One of these questions is the question of the taste of the product.

It is very difficult for someone to rate the taste diversity of such popular products as wine and coffee.

It is possible to walk long along the showcases in the Department of alcoholic products among the assortment of thousands of bottles of wine, choosing one of them for dinner. The same goes for coffee. Its taste strongly depends on the variety, the region in which it grew, the process of cultivation and roasting of coffee beans and, finally, the way of brewing. Here you can trust the advice of sommeliers and baristas, but even they in the evaluation of taste are oriented to subjective parameters, their choice may not be at all what you would like most.

There is a need to develop objective criteria for evaluating taste.

Such criteria can be developed using chemometrical methods.

Taste peculiarities are caused by the chemical composition of the product
It is possible to investigate the chemical composition of a product by methods of optical spectroscopy. The total contribution of interaction of product molecules with optical radiation leaves unique fingerprints in the optical spectrum. The analysis of this optical spectra gives information about taste characteristics.

Problem statement:

A dataset of optical spectra and photographs is formed for some set of wines and different types of coffee. These spectra and photographs are compared with the types of wine and coffee, the level of sweetness, acidity and extraction obtained by objective measurement methods.

On the basis of this data to build a model capable of classifying wines, coffee, and a model capable of distinguishing sweetness, acidity of wine and the level of extraction of coffee.

Forecasting of road accidents in Moscow
Weather conditions and their abrupt changes affect the number of accidents. In bad weather (during downpours, snowfalls, abrupt changes in temperature) the number of occurrences naturally increases. But the significant influence is not only obvious factors. For example, changes in atmospheric pressure reduce the level of attention of road users, as a result, the number of accidents increases significantly.

CODD of Moscow provided unique data on accidents in 2013-2017 GG. The information is collected throughout Moscow, including new Moscow and it has about 2 million events. The statistics on accidents with human victims and material damage are separately reflected, each record contains the date and time of the incident.

You have to use machine learning technologies provided CODD up-to-date data on accidents, as well as open meteorological sources, to build a predictive model of the number of accidents in Moscow depending on the calendar day, Time of the year, hour of the day and weather conditions.

Identification of personality type by photo
There are many typologies of personality. One of the most popular is socionics based on Jung's theory, where the types are defined based on four dichotomies (mutually exclusive properties).

The dichotomy in Socionics is:

− Logic – ethics,

− Intuition – sensor,

− extroversion – introversion,

− Rationality – irrationality.

Each person can be compared only to one dichotomy from a pair, thus, having received 16 possible types of the person.

More information about socionics and types can be read here..

Depending on the type, people have different strengths and weaknesses, you can predict their preferences and characteristic features of behavior. Therefore Socionics is actively applied in personnel consulting, career guidance, business and etc. Well-known companies use or plan to use psychotypes of users to customize advertising and customer service.

Socionic type is defined in the course of a detailed interview, and the specialists pay special attention to the appearance of the person and his non-verbal manifestations. There is a hypothesis that separate dichotomies and socionic type of the person with sufficient accuracy can be assumed on a set of his photos. The problem of building predictive models is the lack of verified data sets about the types of people.

You will be offered lists of people, about whom it is reliably known to what type they belong and possessing the most characteristic features. Analyzing their photos, you have to build a model, predicting the photo as separate dichotomies, and types entirely.

Translation from Sign Language
At least 1% of the population of our planet only according to official data has problems with hearing, comparable with the inability to use verbal symbols in communication.

You are presented with a base of photos of 10 different gestures depicted on them.

Statement of the problem.
Using available data, build models that can recognize sign language in real time

Subscribe to our channel in telegram and be aware of all the information about the Hackathon (and not only)!

© All Right Reserved. YRP MSU.
Made on