The purpose of this readme is to explain the code and structure in completing the assessment for the first part of the data science course. This readme outlines the data, packages and codes used and guides the reader as to how each question of the assessment was answered. Each question is answered in a separate project.
To get started, the different projects was created for question 1, question 2 and question 3. The following code was used:
This question addresses some claims a friend made regarding movie critics that perfectly predict films’ popularity and profitability amongst audiences. This was not the case during the mid 2000s. This question tests some of the friend’s theories by addressing three of the statements she made.
The following packages were used:
pacman::p_load(tidyverse, ggthemes)The data for each question has been unzipped and put under the data folder of each question folder. The data for the first question consists of a csv file. This file contains information on the name of various movies, the year during which the movie was released, each movie’s production studio, its audience score and Rotten Tomatoes score, as well as the profitability of the movie. The data was loaded as follows:
MovieData <- read.csv("Question1/data/Movies.csv")The first statement my friend made is: “I firmly remember that Rotten Tomatoes was always a great review platform - and if a movie had a rating of more than 80% on Rotten Tomatoes, audiences would rate it above 85% every time.”
To address this point, I drew a graph that indicates the ratings audiences gave movies that had a rating of more than 80% on Rotten Tomatoes. This was done by creating a function called RottenT_plot that draws the graph using only the movies that had a Rotten Tomatoes score above 80%. This function is stored in the code folder of the question 1 folder. This function was then sourced in and the relevant input was provided to produce the graph.
The movies were grouped into their relative genres to make it easier to see what the rating of audiences were. This graph indicates that animation movies that had a rating of more than 80% on Rotten Tomatoes, had audience scores above 85%. Therefore, for animation movies, my friend’s statement is true. However, for all the other genres my friend’s statement is not true since the audience scores were below 85%. Therefore, my friend’s statement regarding movies having high audience scores if they have a high Rotten Tomatoes score is incorrect.
source("Question1/code/RottenT_plot.R")
graph_1 <- RottenT_plot(MovieData, xaxis_size = 5, xaxis_rows = 3)
graph_1
The second statement my friend made is: “Disney films may not have the highest grossing numbers, but they’ve always been the most profitable of all the leading studios.”
To test this statement, a graph containing the profitability of all the leading studios were drawn to compare the profitability across studios. This was done by creating a function called Profitability_plot that draws a bar plot. NA values for the leading studios are dropped. This function is stored in the code folder of the question 1 folder. This function was then sourced in and the relevant input was provided to produce the graph.
source("Question1/code/Profitability_plot.R")
graph_2 <- Profitability_plot(MovieData, xaxis_size = 10, xaxis_rows = 10)
graph_2
This graph indicates that Independent was the most profitable studio between 2007 and 2011, as it had the highest profitability (just under 120%). Disney’s profitability was around 40% which is much lower than that of Independent. Therefore, my friend’s second statement is incorrect.
The third statement my friend made is: “Audiences are always drawn to the highest grossing films. In fact, I bet the correlation between the world wide grossing numbers and audience scores would be near 80%.”
To test this statement, the correlation between the world wide grossing numbers and audience scores was calculated. The code for this is shown below.
Corr <-
MovieData %>%
summarise(Corr = cor(Worldwide.Gross, Audience..score.., method = "spearman", use = "pairwise.complete.obs"))The correlation between world wide grossing numbers and audience scores is 29.1%, which is far from 80%. My friend’s third statement is therefore incorrect.
To answer this question, interesting results using the recently released Forbes Billionaires list is presented.
The data for the second question consists of multiple csv files. Each file contains information on a specific country. It indicates the name and age of the billionaires residing in that country, their net worth, their rank on the Forbes Billionaires list and the industry in which the billionaire operates.
Since multiple csv files needs to be loaded, a function is created to collate and combine all the files. This, however, was unsuccessful as an error message was provided:
Error: Can’t combine
AgeandAge.
I then searched for a solution to this problem and found this answer:
I therefore added the mutate to change Age to a character variable. However, another error popped up:
Error in mutate(Age = as.character(Age)) : object ‘Age’ not found
I realised that this has to do with the fact that the dataset contain ‘Age’ is not loaded into the function. I therefore tried to insert the data in the function so that it can pick the Age variable up. I did this by defining “Datroot =”Question2/data/"" as the function’s input. However, this still did not work as I received the following error message:
Error in UseMethod(“mutate”) : no applicable method for ‘mutate’ applied to an object of class “character”
I tried to play around with where I placed the mutate and also tried to search for a solution I could not find one.
The data was loaded as follows (this is commented out to prevent the Readme from not knitting):
#source("Question2/code/Data_Collating.R")
#Forbesdata <- Data_Collating(Datroot = "Question2/data/")Due to these errors, I decided to still write out my thinking, even though I don’t have the code available to create the graphs. The first graph I would have plotted would be one comparing the number of billionaires across each country. It would be interesting to see how many billionaires there are in developing vs developed countries. For this graph I would use a bar plot for each country on the x axis and on the y axis I would indicate the number of billionaires listed in that country. I would therefore group the results by Country.
The second graph I would have plotted would be to compare the average net worth of billionaires under the age of 30, between 30 and 40, between 40 and 50, and over the age of 50. I would think that older billionaires have a higher net worth than younger billionaires, but in this day and age there are multiple billionaires under 30 so it could be that my thinking is wrong. For this, I would need to convert the net worth variable to numeric. I would use: “mutate(NetWorth = as.numeric(NetWorth))” to do this. Regarding the graph, I would again use a bar graph with each bar representing the mean net worth of billionaires in the specific age group. On the y axis the net worth will be displayed and on the x axis the different age groups will be displayed. I will therefore have to create new columns using mutate where I group the people accoridng to their age in the different age brackets.
For this question, the answers have been presented in a Texevier pdf.
The data consists of three rds files and one csv file. It is loaded below:
BBCData <- read_rds("Question3/data/tweets_bbc.rds")
CNNData <- read_rds("Question3/data/tweets_cnn.rds")
EconData <- read_rds("Question3/data/tweets_eco.rds")
CountryData <- read_csv("Question3/data/Country_list.csv")##
## -- Column specification --------------------------------------------------------
## cols(
## .default = col_character()
## )
## i Use `spec()` for the full column specifications.