Predicting Movie Rental Durations

Project Overview

This project aims to help a DVD rental company predict how many days a customer will rent a DVD. The objective is to build a regression model that achieves a Mean Squared Error (MSE) of 3 or less. Such a model will assist the company in optimizing inventory management and rental planning.

Dataset

The dataset rental_info.csv contains various features related to DVD rentals.

Methodology

1. Data Preprocessing

The first step involves preprocessing the data by creating new features and removing unnecessary columns:

• rental_length_days: Calculated as the difference between rental_date and return_date, representing the number of days a DVD was rented.
• Special Features Encoding: Binary columns for whether the DVD includes deleted scenes or behind-the-scenes content.
• Feature Removal: Columns such as special_features, rental_date, and return_date are dropped as they are no longer needed.

2. Feature Selection with Lasso Regression

Lasso regression is used for feature selection, which helps identify the most relevant features by penalizing less important ones:

• Lasso applies a regularization term that forces some feature coefficients to zero, effectively removing them.
• After applying Lasso, the remaining features are used to train a Linear Regression model.

3. Model Training

Two different regression models were explored:

• Linear Regression: This is a basic regression model trained using the features selected by Lasso. It is evaluated on the test set to check its performance.
• Random Forest Regressor: A more complex ensemble method, which uses multiple decision trees to make predictions. A RandomizedSearchCV is applied to find the best hyperparameters for the model, such as the number of trees (n_estimators) and the maximum depth of the trees (max_depth).

4. Evaluation Metrics

The models are evaluated based on their Mean Squared Error (MSE), which measures the average squared difference between the predicted and actual values. The goal is to minimize this error, aiming for an MSE of 3 or less.

Conclusion

The Random Forest model, tuned with optimal hyperparameters, provided the best performance with an MSE of approximately X. This model will help the DVD rental company make better predictions about rental durations, allowing for more efficient inventory management.