AN EMPIRICAL COMPARISON OF MACHINE LEARNING MODELS FOR TIME SERIES FORECASTING

Time series data analysis and forecasting stands as a critical information source, shaping future decision-making, strategy formulation, and operational planning across diverse industries. Ranging from marketing and finance to education, healthcare, and robotics, the time series data has become pivotal in guiding effective actions. Time series data analysis plays a pivotal role in understanding sequential trends and patterns present in the data. The Time series forecasting has been used for prediction for effective decision making. The forecasting techniques consist of statistical models and machine learning models. This paper examines different methods, including AR, MA, ARMA, ARIMA, SARIMA, ARIMAX, SARIMAX, Prophet and LSTM. Two meteorological datasets have been analyzed and the above models have been applied and evaluated using various performance metrics.