The Phenomenon of Temperature Increase in Poland: A Machine Learning Approach to Understanding Patterns and Projections

Abstract

This study presents an analysis of patterns in mean monthly air temperature increases in Poland using the deep learning model Neural Basis Expansion Analysis for Time Series (N-BEATS) algorithm. The dataset comprises mean monthly temperatures recorded between 1951 and 2024 at eight meteorological stations across Poland. The research was conducted in two phases. In the first phase, the 74-year period was divided into two distinct intervals: one characterized by relative temperature stability, and the other by a marked upward trend. In the second phase, the N-BEATS neural network was employed to extract temporal patterns directly from the data and to forecast future temperature values. The results confirm the capacity of machine learning methods to identify persistent climate trends and demonstrate their utility for long-term monitoring and prediction.