ROMIR at the International Statistical Forum of the CIS

04 October 2024

The International Statistical Forum of the CIS is taking place in Tashkent, the capital of Uzbekistan, bringing together leading experts in the field of social sciences. Olga Perfilieva, Director of Client Relations at ROMIR, spoke as an expert with a report on the correct use of big data and artificial intelligence in Human studies.

In her speech, the expert noted that the development of technologies affects all areas of activity, including the marketing research industry. The advantages of big data are already widely known, but have been very little tested and studied in terms of their usefulness and effectiveness.

“Big data increases the level of knowledge about human life, the motives underlying their consumption and choices. Which, in turn, increases the level of customer engagement, brand recognition, and improves business efficiency as a whole,” Olga Perfilieva noted.

ROMIR is always at the forefront of technological development and works with big data, using a comprehensive understanding of Humans from sociological, anthropological and psychological points of view.

In order to comprehensively study a person, ROMIR has created and is developing its own unique Longitudinal System, which collects its own big data about a person throughout the entire history of the Institute.

The vast majority of global Longitudinal Systems are aimed at studying population groups that are exposed to risks. "We decided not to abandon international experience and took the key principles - collecting and analyzing statistics, as well as an element of predictiveness. However, the key is the fact that we do not want to focus on a specific group of people, but build a complete model of modern Russian society," the expert explained.

The ROMIR Longitudinal System operationalizes the concept of a person by more than 300 parameters, segments data about a person, accumulates the integration of panel data with external data about a person of external origin.

"The ROMIR Longitudinal System is a data lake, with the help of which it is possible to normalize the results of any surveys, find and analyze the relationships between different databases through a system of interconnectors. An important milestone of our system is the module for normalizing survey data and harmonizing external data. We have learned to measure people with people. Access to data is optimized for quick receipt of any results. We also work on our BI system, which is based on all the experience we have, as well as the wishes of our clients,” explained Olga Perfilieva.

One of the unique results of ROMIR's work is the normalization of survey data. “We live in an era of crisis in methods for studying people's opinions, such as CAWI, CATI and CAPI. We have found a method for restoring the representativeness of survey data through the interconnector method. For each survey, we take a control group on the Longitudinal System, about which we know absolutely everything. To restore representativeness, we extrapolate the results of the control group on secondary parameters (interconnectors) of human behavior to the main respondents, which leads to the formation of a result relevant to the general population,” said Olga Perfilieva.

ROMIR is currently moving towards predictive analytics, which will allow modeling certain events and assessing the potential reaction of a real person.

In the context of the practical application of artificial intelligence (AI), ROMIR is introducing new methods of collecting and processing data. Today, in connection with the introduction of a survey system via web interfaces, it is becoming possible to use AI at the stage of consolidated processing of the received data.

Today, routine processes of collecting and processing data are associated with the use of new technologies due to the implemented:

systems for integrating and working with the new fiscal data system from the Federal Tax Service (electronic checks), which allows taking into account absolutely all user checks;
machine learning for the coding system (automatic text recognition and automatic categorization).
However, testing several hypotheses allowed us to conclude that artificial intelligence or neural networks are not suitable for large volumes of numerical data. The tasks go beyond the capabilities of the models. Implausible results are generated, which is especially critical in a business context, where the accuracy and reliability of information are important.

ROMIR's own development and approaches are universal in the context of big data, revealing hidden patterns and trends. They allow developing models that can be customized in accordance with unique requirements and data structure.

“The more specific the market, the less likely the artificial intelligence hypotheses are. Artificial intelligence gives superficial hypotheses, in the study of which we find more important and deep matters,” Olga Perfilieva noted.

Applied use of AI for field research tasks, independently


Поделиться в соцсетях:

Подписаться на рассылку: