Kazakhstan to use AI for statistical data processing and analysis

Starting next year, Kazakhstan will use artificial intelligence to collect statistical data. Neural networks will be applied to process fiscal receipts, helping statisticians calculate food prices and analyze citizens’ actual spending. The system automatically identifies product names by code and sorts them into categories, allowing coverage of as many retail outlets as possible. Previously, data was collected manually from 12,000 stores; now it will cover 124,000. The mechanism, tested this year, proved effective: over 20 billion receipts were analyzed, eight billion using AI. The domestic neural network performs calculations faster and more accurately.

«For example, there is 2.5% pasteurized milk, but not all receipts list the full product details. It may appear as ‘pasteurized milk 2.5%’, or sometimes just ‘milk’, or only the percentage. Such inconsistent data cannot be extracted and used directly. Therefore, the artificial intelligence system is extensively trained to clean the data, providing statisticians with more accurate and processed information,» noted Kuanyshbek Zhakypbekuly, Director of the Price Statistics Department at the Bureau of National Statistics.

«For some, and indeed many, indicators, data processing used to take a very long time. For example, processing information on GDP could take up to 20 days. Now, thanks to AI, all of this has been streamlined, and processing takes less than an hour,» said Maksat Turlybayev, Head of the Bureau of National Statistics, Kazakh Agency for Strategic Planning and Reforms.

Web-editor: Арайлым Серқанова