Empowering Decision-Makers: Integrated Approach for Anomaly Detection and Temperature Forecasting

Abstract

This research offers an original approach to anomaly identification and temperature forecasting by combining LSTM neural networks with ARIMA models. Our goal is to estimate temperature fluctuations using ARIMA models. Then, we use the forecasts as inputs for anomaly detection based on LSTM, which identifies patterns that don't match expectations. We review relevant literature on temperature forecasting, emphasizing the role of order selection and external factors. Significant proficiency in identifying temporal connections is shown by LSTM networks in anomaly detection. Our integrated strategy enhances the detection of anomalous temperature events by bridging the gap between anomaly detection and forecasting. This research gives decision- makers in all industries affected by anomalies and temperature variations more authority.