This research chapter delves into the domain of suggestion mining within game reviews based on sentiment analysis. We explored the efficacy of employing pre-trained sentiment analysis models in enhancing accuracy for suggestion mining tasks. Specifically, we investigate the performance of customizing a pre-trained DRC_Net model for suggestion mining in game reviews compared to using a generic pre-trained model like DistilBERT. Through a comparative study, we highlight the importance of transfer learning in this context, leveraging the knowledge embedded within pre-trained models trained on broader sentiment analysis tasks. By fine-tuning these models on our domain-specific game review dataset, we successfully transferred learned representations and knowledge to our specialized task of suggestion mining. Our findings demonstrate that the customized DRC_Net pre-trained model outperforms the generic pre-trained DistilBERT model and previous suggestion mining efforts using the DRC_Net model. This study underscores the practical significance of transfer learning in optimizing suggestion mining tasks utilizing sentiment analysis within domain-specific contexts.