Introduction

View Github Repository Open presentation in a new window

julienfbeaulieu

See all presentation from julienfbeaulieu

Introduction

0 0

oscarSlides

On Github julienfbeaulieu / oscarSlides

Introduction

  • Every year, Academy Awards are given to the best movies featured in the previous year.
  • The Oscar shiny app explores data from the movies that were nominated in the Oscar ceremonies.
  • The app can assessed on the shinyapps.io
  • Alternatively, the app can be run locally by:
  • downloading the server.R and ui.R files from the github,
  • setting the working directory to where the files are located,
  • running the following R code:
library(shiny)
runApp()

Methodology

The original dataset lists information about each nomination for each year.

  • Since an award can either go to a film or to an artist, I processed the database to distinguish the two. The resulting oscar.csv database lists 10 variables about 10098 nominations.

From this, a second dataset was created and takes the point of view of each film. Number of academy nominations and awards were summed, and extra information was pulled out using the omdb api. The resulting movie.csv database lists 27 variables about 4361 nominated movies.

  • For instance, the following call retrieves information about the titanic") movie:
  t <- "Titanic"
  y <- "1997"
  url <- paste0("http://www.omdbapi.com/?t=",t,"&y=",y,"&plot=short&r=json")

  temp <- as.data.frame(fromJSON(url))

Side panel

Filter movies

In the side panel, you can filter the dataset, either by year, by number of oscar nominations or awards. - For instance, the data can be filtered for the period 1980-2000, with a minimum number of 5 nominations and 0 awards. - The movie database is the filtered using the following r code:

    minYear <- 1980; maxYear <- 2000; minNomination <- 5; minAward <- 0
    movieSubset <- filter(movie, Year >= minYear, Year <= maxYear, 
                          Academy.Nominations >= minNomination, Academy.Awards >= minAward)

Select variables

Additionnaly, you can select variables for the x- and y-axis of the graph and the for the linear model. - For instance, the metascore and the imdb user rating could be selected as the x and y variables.

    plotData <- data.frame(y = movieSubset[["imdbRating"]], x = movieSubset[["Metascore"]])

Output graph

For the preceeding filtered subset of movies and the variables selected, the Oscar shiny app would output the following graph:

Output linear model

  • For the same filter and variable setting, the Oscar shiny app would ouput this linear model:
## 
## Call:
## lm(formula = as.formula(paste("imdbRating", " ~ ", "Metascore")), 
##     data = movieSubset)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.05376 -0.39359 -0.09442  0.34108  1.50724 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 6.166080   0.432039  14.272  < 2e-16 ***
## Metascore   0.020333   0.005502   3.696 0.000421 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5305 on 73 degrees of freedom
##   (46 observations deleted due to missingness)
## Multiple R-squared:  0.1576, Adjusted R-squared:  0.1461 
## F-statistic: 13.66 on 1 and 73 DF,  p-value: 0.000421
  • Namely, for the highly nominated movies (>= 5 nominations) of the period 1980-2000, the Metascore is highly significant in predicting the imdbRating. Though, the correlation between the two variables is really low.