Research Paper On: A Movie
Recommendation System Using
Supervised Learning Algorithms
Abstract
In the age of digital entertainment, personalized content delivery has become essential for user
engagement. Recommendation systems serve this need by suggesting relevant items—movies,
in this context—based on user preferences. This research focuses on building a movie
recommendation system using supervised machine learning techniques. While collaborative
filtering and deep learning have been widely explored, this paper investigates the effectiveness
of supervised algorithms like Decision Trees, Random Forests, and Support Vector Machines
(SVMs) in classifying and predicting movie preferences based on historical data. The goal is to
build a model that learns from user ratings and features of movies to predict whether a user
would like a given movie.
1. Introduction
With the exponential growth of online streaming platforms, users are overwhelmed with
content. Movie recommendation systems help filter this vast array of options. While
unsupervised techniques such as collaborative filtering dominate this field, supervised learning
algorithms provide an alternative path by treating the recommendation task as a classification
or regression problem. This paper proposes a recommendation system using supervised
learning, where user preferences are predicted based on labeled training data.
2. Background and Related Work
Recommendation systems are broadly categorized into:
• Content-Based Filtering: Recommends items similar to those the user has liked.
• Collaborative Filtering: Recommends items liked by similar users.
• Hybrid Systems: Combines the two above.
In supervised learning, models are trained on labeled data. Prior studies using Decision Trees,
Logistic Regression, and Neural Networks have shown promise in predicting ratings and
preferences, especially when enriched with feature engineering.
3. Problem Statement
The objective is to develop a machine learning model that predicts whether a user will like a
movie based on features such as genre, cast, director, and previous user ratings. This is treated
as a binary classification problem (like/dislike).
4. Dataset and Preprocessing
4.1 Dataset
The MovieLens dataset is utilized, containing:
• User ratings (0.5 to 5 stars)
• Movie metadata (genres, titles, release year)
• User demographics (optional)
4.2 Preprocessing
• Label Generation: Ratings ≥ 3.5 are labeled as “like” (1), others as “dislike” (0).
• Feature Engineering:
• One-hot encoding for genres
• TF-IDF for movie descriptions (if available)
• Encoding user IDs and movie IDs
• Train-Test Split: 80-20 ratio
5. Methodology
5.1 Supervised Algorithms Used
• Decision Tree Classifier: Simple and interpretable
• Random Forest Classifier: Ensemble technique that improves accuracy
• Support Vector Machine (SVM): Effective in high-dimensional spaces
5.2 Evaluation Metrics
• Accuracy: Correct predictions / Total predictions
• Precision, Recall, F1-Score: For measuring recommendation quality
• Confusion Matrix: To analyze prediction distribution
6. Results and Discussion
MODEL ACCURACY PRECISION RECALL F1-SCORE ROC-AUC
Decision Tree 72.4% 0.70 0.73 0.71 0.75
Random Forest 80.2% 0.81 0.79 0.80 0.87
SVM(RBF Kernel) 76.1% 0.74 0.75 0.74 0.81
Random Forest outperformed other models due to its robustness and ability to reduce
overfitting. SVM also showed good performance, particularly with tuned hyperparameters.
7. Conclusion
Supervised learning algorithms can be effectively applied to movie recommendation systems.
While not as commonly used as collaborative filtering methods, supervised models—
particularly ensemble models like Random Forest—can provide high accuracy and flexibility
when combined with rich feature sets. Future improvements include using user behavioral data
and incorporating deep learning for feature extraction.
8. Future Work
• Integrating user feedback loops for continual learning
• Using deep learning models like neural collaborative filtering
• Enhancing with NLP-based sentiment analysis on user reviews
References
1. Harper, F. M., & Konstan, J. A. (2015). The MovieLens Datasets: History and
Context. ACM Transactions on Interactive Intelligent Systems (TiiS).
2. Aggarwal, C. C. (2016). Recommender Systems: The Textbook. Springer.
3. Breiman, L. (2001). Random Forests. Machine Learning, 45(1), 5–32.
