The high performance for optimization in different KPIs is a highly important feature in the LTE network, as customer satisfaction depends on it directly. This work describes advanced feature engineering and the Random Forest regression model to predict customer satisfaction based on the use of network KPIs. Herein, a novel model is proposed, based on the Random Forest algorithm with features such as Signal Stability, Quality-Throughput Interaction, and a Satisfaction Index, to catch complex relationships between traditional KPIs. The performance of the proposed model was characterized by high predictive accuracy, seen by an R-squared of 0.87, which denotes a high capability of explaining customer satisfaction variance. The most important variables featured in this feature importance analysis were indeed Signal Stability and Quality-Throughput Interaction, which underlined the fact that network operators had to focus both on signal quality and data throughput. These findings provide insights for network performance improvement and user satisfaction by showing how much feature engineering combined with predictive modelling can help enhance LTE network performance.