Forecasting Weekend Box-Office Movie Revenue

November 22, 2017

Report Abuse
This collection is comprised of a tutorial and experiments that form a solution to the problem of predicting weekly revenue for movies at the box-office.
Envision a use case involving a marketing team for a movie studio that wants to do targeted advertising each week before the weekend at the box-office. In order to do so, they must identify movies that they believe will underperform in the upcoming weekend. The team therefore needs revenue predictions each week for all the movies at the box-office for the upcoming weekend. Before the weekend begins, they will use these predictions in order to formulate their targeted marketing campaigns. Here, we focus on the movie revenue predictions. Using data from [][1], the contents of this collection are focused on building a machine learning model that can predict movie revenue for the top X movies at the box-office for a given weekend. The following items are contained in the collection: 1. **[Tutorial][2]**: A link to a set of three Jupyter Notebooks that outline the process of scraping data, engineering features, building a machine learning model, deploying the model, analyzing the results, and leveraging the deployed web-service to make predictions. 2. **[Experiment][3]**: Feature correlations to the outcome (movie gross / revenue) for the dataset. 3. **[Experiment][4]**: A least squares regression model for predicting movie revenue for a given weekend. Training and statistical validation of the model are included. 4. **[Experiment][5]**: A predictive experiment for the least squares regression model that can be deployed as a web-service to make revenue predictions for a single movie on a single weekend. [1]: [2]: [3]: [4]: [5]: