MASHINALI O‘QITISH ALGORITMLARI YORDAMIDA BOSH MIYA O‘SMALARINI ANIQLASH MODELI

Аннотация

The article develops a model of multi-feature linear regression, one of the algorithms of machine learning with a teacher, for forecasting inflation processes in Uzbekistan. Estimating the inflation rate based only on the time (year) factor is a very limited approach and ignores the main factors affecting economic changes. Therefore, several factors that significantly affect inflation were introduced into the model: the consumer price index (CPI), the wage growth rate, the unemployment rate and the exchange rate. The model used an artificial inflation_multi_feature.csv file created based on historical data. The R² coefficient in the trained model showed a value of 0.847, which means that it can explain 84.7 percent of inflation changes. The MSE (root mean square error) was only 1.48. These results show that the regression model built taking into account several factors has a much higher accuracy. The results of the study prove that when forecasting inflation, it is necessary to analyze indicators that comprehensively characterize economic activity, and not rely only on the time parameter. This model can help in public policy, budget planning and the formation of monetary policy in the banking sector. At the next stage, it is planned to integrate time series models into the model and test it on real statistical data. Keywords: Artificial intelligence, economic processes, machine learning, MSE, supervised learning.