This project is designed for film directors' works analysis. What we want to explore is the relationship between films' performance and film directors' background.
-
Directors dataset: Extract directors' birthplace/birthday/bio/trivia/race/gender information
- A short VBA code to automatically download images from url address and insert in Excel:
- Fill in Excel cells by automatically downloading images from url address
-
Scores dataset: Extract Users' ratings and Critic scores/comments for Top films
-
Users ratings distribution for TOP200 films each year starting from 2004:
-
Critic/Metascore ratings distribution for TOP200 films each year starting from 2004:
-
-
Updated info - metacritic contains complete critic info compared with IMDb
-
Nature Language Processing Analysis - Topic modeling and text analysis for comments
- LIWC: LIWC transparent text analysis
- By using LIWC Topic model analysis, we calculate the word frequency of critic rating comments related to specific ten topics. The LIWC Extractor code in notebook can be found here
- LSA/LDA decomposition: A simplified notebook to implement basic LDA/LSA analysis for sample text
- LIWC: LIWC transparent text analysis
-
Word Topic frequency statistical analysis given race and gender info
- We have fixed around 10 topics that we may interest from LIWC. What we did here was to extract word-text related to these topics, gave statistical info and compared how their frequencies may vary among people with different race and gender. Notebook details can be found here
-
Login Web scraping
- Task1: (Login not needed) Use "bomlink" and "imdblink" of each film to scrape information including 'releasedate' and 'closedate' and 'number of critics'.
- Task2: Here we want to scrap StarPower values for the Top5 casts for each film measured 10 weeks prior to the release date (scraped from Task1). To be more specific, we would use Selenium to set up a driver and log in our account first. And by visiting Pro-IMDb profile of casts, we would tell certain period rankings through graph Data automatically.
- Task3: Find out each film's release date and match the date to week's gross and total number of movies info identified by boxofficemojo.com
- " Season captures whether a distributor released a film during a high-demand period. Though most peak seasons revolve around holidays, we developed a continuous measure of seasonality by creating a moving average of total box office receipts across all films for a three-week window centered around the film’s release date during the previous three years" (Sorenson and Waguespack, 2006: p. 572)
-
unconventionality measures (% of all prior films with same first 2 descriptors in imdb)
- Own unconventionality: all prior own films during the past 5 years.
- Unconventionality compared to others: all prior films during the past 5 years.
- Notebook for own unconventionality details can be found here and collection results uploaded
- Notebook for unconventionality compared to others details can be found here and collection results uploaded and feature films genre collection from year 2000 to year 2015
-
collect top movies' IMDb descriptors listed in BoxOfficeMojo from year 2000 to 2003: Notebook code and collection results can be found here
-
Collected some featured movies and document the awards/nominations that directors in each movie has won.Notebook code and collection result can be found here.
-
Collected featured movies' exhaustive genres lists based on BoxOfficeMojo websites. Notebook code and collection result can be found here. Also, (login needed) collected featured movies' exhaustive genres lists based on IMDb Pro websites. Check PyCharm code files and collection results.
