HSE Students Take First Place in Kaggle International Data Analysis Competition
Artyom Volgin and Ekaterina Melianova, second-year students of the Master’s programme in Applied Statistics with Network Analysis, outperformed more than 100 teams from different countries to take first place in the DS4G: Environmental Insights Explorer Competition hosted by Kaggle.
In the competition, participants had to develop a methodology for using satellite data to calculate the intensity of nitric oxide emissions. Artyom and Ekaterina analyzed meteorological data, power plant information, and emissions statistics from the past year in Puerto Rico.
‘Essentially, we had to determine how environmentally friendly the production of electrical power is on the island and then build a model by which we could calculate the intensity of emissions for other areas,’ Artyom explains. ‘The task was simple, but it took us a few days to realize that.’
According to the students, they had never worked with satellite data and only knew a bit about power grids before the competition. ‘Also, Puerto Rico is a complicated area for experiments like this,’ Artyom adds. ‘Electrical power plants on the island are pretty weak, and their effect is difficult to distinguish from the rest of the nitric oxide pollution in the air, which is mainly emitted by automobiles.’
The competition lasted about six weeks, and more than a hundred teams from India, Spain, France, and other countries participated. For Artyom and Ekaterina, this was their second time competing in a Kaggle data analysis competition—last year they took second place. This time they not only took first place but took home a cash prize of $10,000.
Kaggle is a data science platform of Google. The community brings together about 3 million specialists in data processing and machine learning all over the world. The resource publishes training materials, organizes surveys, and holds online competitions. In the DS4G: Environmental Insights Explorer Competition, participants solve different applied problems, and expert judges evaluate their work in accordance with criteria such as the quality of their models, the informational value of their visualizations, the usefulness of their recommendations, and so on.
The students noted that skills they gained at HSE, such as formulating substantive problems using static methods, came in handy during the competition.
Our knowledge of a wide range of applied analytical tools, which we gained in our master’s programme, helped us choose an appropriate model for the task
When asked what recommendations he would give to other students who wish to participate in data analysis competitions, Artyom says that he would suggest working on your research skills. It is important to be able to formulate questions, determine appropriate methods for finding a solution, articulate the importance of the work, process and analyze data, and interpret the results.
‘The ability to tell a story plays a particularly important role,’ says Artyom. ‘You need to be able to clearly and succinctly explain your train of thought with the aid of visualizations. We mainly learn how to do this while working on research projects (such as research papers or dissertations), so it’s important to really give these types of assignments and projects serious attention. It’s also important to have a handle on a wide range of applied data analysis methods and be willing to study them in depth while you’re working.’