With an Accuracy of 93%: AI Taught to Predict Inflation Based on News

The system makes three forecasts — for a month, a quarter, and a year

At the North Caucasus Federal University (NCFU), artificial intelligence (AI) has been developed to predict inflation based on news data. This was reported by the university's press service.

 The resulting system helps to make three inflation forecasts: short-term (for a month), medium-term (for a quarter), and long-term (for a year).
Press Service of NCFU

To create the model, scientists collected and labeled data, using scripts for automated article collection and manual classification by key parameters. This allowed for the formation of a balanced dataset and helped to avoid bias in the model. The new analysis system demonstrated high classification accuracy — 93%.

Our project has confirmed the effectiveness of using machine learning for economic analysis. In the future, we plan to regularly update the data and use more complex architectures.
Mikhail Babenko, Project Manager

In the future, such a system may be useful for business representatives.

Earlier, www1.ru reported that in Moscow a system was developed to predict stock price growth.

Read more on the topic:

MIPT Taught Artificial Intelligence to Write Prescriptions and Sick Leave Certificates

Will Solve All Tasks, Answer All Questions: A Multipurpose Chatbot Based on AI Developed in Buryatia

The Ministry of Digital Affairs Launched Testing of Generative AI on "Gosuslugi"