A vast amount of films are made in the current age of digital entertainment, but the challenge is in sorting through the millions of pieces of material that are generated annually to identify the needed content. Recommendation engines, however, are much more useful in these circumstances. In the streaming platform, these systems empower users to discover new and relevant movies, benefiting both viewers and the entertainment industry. Collaborative filtering, also known as content-based filtering, is one of the most often used techniques in recommender systems to forecast things that a user would find interesting. The purpose of this work is to enhance the performance and accuracy of a regular filtering approach by using association rule mining and ML-based hybrid collaborative content filtering. In this study, to build an effective movie recommendation system used the MovieLens 1 M dataset from an open-source Kaggle website. To train and test the model this data has been split into an 80:20 ratio. A hybrid recommendation model is created using the dot product of user and movie embeddings, attaining an accuracy71.85%, precision74%, recall78.42%, and an F1-Score76.15%. Furthermore, similarity measures such as Cosine (0.4225), Pearson (0.5076), Jaccard (0.5076), Euclidean (0.5076), and Manhattan (0.3856) are used to evaluate the associations between users and movies. Also, the study uses rule mining with the Eclat algorithm to improve suggestions. This algorithm finds common groups of items and makes association rules that show hidden connections between users and movies. The comparative analysis shows the proposed hybrid model perform well in compare to other deep learning and Explainable Matrix Factorization (EMF) models in terms of RMSE measure. The findings confirm a reliable and tailored movie recommendation system that exhibits exceptional performance across many assessment measures.