Optimized Movie Recommendation via Sentiment Analysis & Hyperparameter Tuning
Manisha Valera, Rahul Mehta · Jurnal Kejuruteraan · 2025
For improving movie recommendation schemes, this study is targeting to overcome dilemmas like data sparsity and cold start issues, which can limit the relevance and accuracy of recommendations. Progressive tactics are introduced by merging Count Vectorization, Cosine Similarity, Truncated SVD, Linear SVC for sentiment analysis, and Linear Regression for rating prediction, whose objective is to tackle the complications faced by conventional movie recommendation systems, finally gaining superior accuracy as well as finesse in both recommendation and sentiment analysis tasks. In this paper, it explores the challenges of hyperparameter tuning in machine learning, particularly the shortfalls of traditional methods -Grid Search and Randomized Search. Here it inspects progressive techniques as well, such as Bayesian optimization and Optuna, which improve model performance by optimizing hyperparameters more proficiently and reducing computational costs. The proposed methodology governs remarkable effectiveness in overcoming data sparsity and cold start anomalies, plus achieving superior accuracy in recommendations and finesse in sentiment analysis. by achieving an accuracy of 99.87%. scores, The Optuna-optimized LinearSVC model reveals outstanding sentiment classification abilities. at macro and micro levels, this performance is highlighted by its flawless precision, recall, and F1-score, which highlights its accuracy and balanced classification across all classes. Same way, showcasing robust regression performance, the Optuna-optimized LinearSVR model attained a Mean Absolute Error (MAE) of 0.596 and a Mean Squared Error (MSE) of 0.7245. These metrics indicate that the model provides accurate continuous predictions with minimal error. This study makes a noteworthy growth in movie recommendation systems, promising notable evolutions in accuracy and sophistication, flagging the way for future developments in the field.