REGRESSION AND TIME SERIES ANALYSIS FOR CLIMATE TRENDS IN KAZAKHSTAN

Abstract

Climate change is an important issue for Kazakhstan due to its diverse geography, which results in varying temperature and precipitation patterns in different regions. This study examines climate trends in Kazakhstan using B-spline smoothing, functional regression, and time series forecasting models, SARIMA and ETS. The analysis focuses mainly on temperature and precipitation data from 2000 to 2024. To begin, B-spline smoothing was applied to refine the data, while functional regression helped to explore deeply the relationship between temperature and precipitation. A correlation matrix revealed positive, negative, and no linear relationships between seven climate factors. The SARIMA and ETS models were then used to predict temperature trends for the next 24 months. The SARIMA model effectively captured seasonal variations, while the ETS model delivered more precise forecasts, reflecting significant seasonal fluctuations. These results provide valuable information on Kazakhstan’s climate, helping to shape strategies for agriculture and sustainable development going forward.