Algeria is a country renowned for its gas exports, but there are concerns that the increasing local demand may jeopardize its position as an exporter. In the face of these challenges, it is crucial to improve methods for predicting natural gas consumption. This article presents an innovative approach that combines long Short-Term memory (LSTM) and Seasonal AutoRegressive Integrated Moving Average (SARIMA) methods for predicting natural gas consumption. An accurate predictive model was developed using this approach, utilizing a dataset provided by SONELGAZ CD Bejaïa, which includes natural gas consumption data for high-pressure clients from 2015 to 2022. The results demonstrated a prediction error rate of 4.85% for individual client consumption and 4.76% for the entire client set. These findings showcase the effectiveness of the approach in accurately predicting future natural gas consumption. By combining LSTM and SARIMA, the approach offers a significant improvement over individual methods. By anticipating demand fluctuations, this approach contributes to effectively managing this valuable resource, enabling optimal utilization of natural gas in Bejaïa.