Comparative Analysis of Time Series Forecasting Models for Weather Prediction: ARIMA vs. STL

This research compares two well-known models for predicting weather: ARIMA (Auto Regressive Integrated Moving Average) and STL (Seasonal-Trend Decomposition using Loess). Accurate weather forecasts are crucial for businesses and daily life, making strong forecasting methods necessary. ARIMA finds linear patterns and time dependencies in data, while STL splits time series into seasonal, trend, and random parts. Using real-world weather data, we train and test these models. We compare their performances using MSE (Mean Squared Error) and R2 (R-squared) metrics. The research demonstrates that the STL model performs better than the ARIMA model at accurately predicting weather patterns and fitting to the data. This superior performance makes the STL model a more dependable choice for weather forecasting. This study enhances techniques for weather prediction and offers valuable information to those making decisions in weather-sensitive businesses.