WEATHER FORECASTING USING TIME SERIES

Abstract

In today’s world, accurate and reliable weather forecasts are crucial for various areas of human life, from agriculture to transportation, the energy sector, and tourism. Accurate weather forecasts allow farmers to plan crop rotation and field cultivation, and help businesses manage energy resources (planning the heating season, fluctuations in energy consumption, etc.). In this context, time series analysis is of particular importance, as it allows you to study and predict changes in key weather indicators (temperature, day length, humidity, etc.). A time series is a sequence of data points that appear in a certain order over a certain period of time [1]. In the context of weather research, time series are data on various weather indicators (temperature, humidity, atmospheric pressure, air direction and speed, etc.) over a certain period of time. Analyzing these series allows you to identify trends, seasonal fluctuations, and other patterns that can be used to predict future values. To solve the problem of weather forecasting, which has a pronounced seasonality, this paper uses the Seasonal Autoregressive Integrated Moving Average (SARIMA) model. SARIMA is a versatile and widely used time series forecasting model. It is an extension of the non-seasonal ARIMA model designed to process data with seasonal fluctuations [2]. The SARIMA model is characterized by the parameters (p, d, q)(P, D, Q)m, where p, d, q are responsible for the non-seasonal part of the model (autoregression, difference and trend moving average, respectively), and P, D, Q are responsible for the seasonal part. The parameter m determines the periodicity, i.e. seasonality (for example, 12 for monthly data with annual seasonality) [2]. The study involved data acquisition and processing, checking the data for signs of stationarity, finding the optimal parameters of the SARIMA model, and evaluating the accuracy of the forecasting results.