We will cover the following tasks in 49 minutes:
Data Loading and Exploration
In this task, we are introduced to the project and learning outcomes. Once we are familiarized with the Rhyme interface, we begin working in Jupyter Notebook, a web-based interactive computational environment for creating notebook documents.
Next, we will import essential libraries such as
matplotlib. Lastly, we will load our data sets into memory using pandas and look at the first few entries.
The TMDB data set contains 7400 movies and a variety of metadata. Movies are labeled with
id. Data points include cast, crew, plot keywords, budget, posters, release dates, languages, production companies, and countries.
Visualizing the Target Distribution
This database was initially released so that teams could treat it like a regression problem and predict the worldwide box office revenue of 4400 movies provided in the test file.
However, we will use this feature rich database for data analysis and data visualization in this project. In the next project, we will use the same database for feature engineering and feature visualization.
To proceed to feature analysis, we first need to visualize the target. Using Seaborn’s
distplot function, we will plot the distribution of movie revenues. To illustrate the skew, we will also plot the distribution of revenue on a logarithmic scale using
Comparing Film Revenue to Budget
Just as we did in Task 2, we will create two subplots. One for the distribution of the
budget; the other to plot the log budget against log revenue. We perform the log transformation to make the distributions more manageable.
Do Official Homepages Impact Revenue?
In this task, we will use the data to demonstrate the effect of an official homepage on movie revenue. We will create a binary feature
has_homepage to indicate the presence or absence of official movie homepages. We can use Seaborn’s
catplot() function to plot the revenues for movies with and without homepages using our newly created features.
Distribution of Lanuages across Films
As Hollywood is known globally, we would expect the data to support our intuition that English movies generate the highest revenue. In this task, we will use box plots to test this hypothesis. We might be surprised that our intuitions about the world may not always align.
Common Words in Film Titles and Descriptions
It might be of interest of to identify trends across movie titles and descriptions and their effect on revenue. In this task, we will generate word cloud for movie titles and descriptions. Word Cloud is a data visualization technique used for representing text data in which the size of each word indicates its frequency or importance.
How do Film Descriptions Impact Revenue?
To identify which words have the highest impact on revenue, we tokenize and vectorize movie titles and descriptions, fit a linear regression model to it, and use ELI5 to display the high impact words.
About the Host (Snehan Kekre)
Snehan hosts Machine Learning and Data Sciences projects at Rhyme. He is in his senior year of university at the Minerva Schools at KGI, studying Computer Science and Artificial Intelligence. When not applying computational and quantitative methods to identify the structures shaping the world around him, he can sometimes be seen trekking in the mountains of Nepal.