Temperature Forecast Prediction and Anomaly Detection

Abstract

By integrating ARIMA models and LSTM neural networks, this study provides a unique method for temperature forecasting and anomaly detection. Our objective is to use ARIMA models to estimate temperature variations. We then use the forecasts as inputs for LSTM-based anomaly detection, which detects patterns that deviate from expectations. We examine pertinent past research in temperature forecasting, highlighting the significance of external influences and order selection. LSTM networks demonstrate significant skills in detecting temporal dependencies in anomaly detection. By bridging the gap between anomaly detection and forecasting, our integrated approach improves the identification of anomalous temperature events. Decision-makers in all industries influenced by anomalies and changes in temperature are empowered by this research.