Project 3
Elina
2025-03-20
Introduction
Movie Market Insights Through Descriptive and Predictive Analytics Using RStudio
As marketing analysts for a movie production company, we are tasked in understanding the movie market, evaluating their box office financial performance and factors impacting it, and recommending ways to increase their revenue.
Research Question
What factors affect box office revenue, and what strategies can be implemented to increase this?
Libraries
library(readxl)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(ggplot2)
library(tidyr)
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ forcats 1.0.0 ✔ readr 2.1.5
## ✔ lubridate 1.9.4 ✔ stringr 1.5.1
## ✔ purrr 1.0.2 ✔ tibble 3.2.1## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorslibrary(data.table)##
## Attaching package: 'data.table'
##
## The following objects are masked from 'package:lubridate':
##
## hour, isoweek, mday, minute, month, quarter, second, wday, week,
## yday, year
##
## The following object is masked from 'package:purrr':
##
## transpose
##
## The following objects are masked from 'package:dplyr':
##
## between, first, lastlibrary(stringr)
library(corrgram)
library(car)## Loading required package: carData
##
## Attaching package: 'car'
##
## The following object is masked from 'package:purrr':
##
## some
##
## The following object is masked from 'package:dplyr':
##
## recodelibrary(randomForest)## randomForest 4.7-1.2
## Type rfNews() to see new features/changes/bug fixes.
##
## Attaching package: 'randomForest'
##
## The following object is masked from 'package:ggplot2':
##
## margin
##
## The following object is masked from 'package:dplyr':
##
## combinelibrary(caret)## Loading required package: lattice
##
## Attaching package: 'lattice'
##
## The following object is masked from 'package:corrgram':
##
## panel.fill
##
##
## Attaching package: 'caret'
##
## The following object is masked from 'package:purrr':
##
## liftCleaning the Data
Cleaning the data, removing N/A values, renaming titles for efficiency, trimming extra space in the title column and removing any extra spaces, and finally merging the datasets together.
Audience<-read.csv("Movie Dataset_General Audience.csv")
View(Audience)
summary(Audience)## original_title type genre runtime
## Length:651 Length:651 Length:651 Min. : 39.0
## Class :character Class :character Class :character 1st Qu.: 92.0
## Mode :character Mode :character Mode :character Median :103.0
## Mean :105.8
## 3rd Qu.:115.8
## Max. :267.0
## NA's :1
## mpaa_rating imdb_rating imdb_num_votes critics_rating
## Length:651 Min. : 1.200 Min. : 180 Length:651
## Class :character 1st Qu.: 4.300 1st Qu.: 4035 Class :character
## Mode :character Median : 6.100 Median : 10651 Mode :character
## Mean : 5.811 Mean : 49378
## 3rd Qu.: 7.100 3rd Qu.: 56156
## Max. :55.000 Max. :893008
##
## critics_score audience_rating audience_score best_pic_nom
## Min. : 5.00 Length:651 Min. :13.00 Length:651
## 1st Qu.: 36.50 Class :character 1st Qu.:38.00 Class :character
## Median : 55.00 Mode :character Median :55.00 Mode :character
## Mean : 54.95 Mean :56.56
## 3rd Qu.: 73.50 3rd Qu.:76.00
## Max. :100.00 Max. :96.00
##
## Facebook_Likes
## Min. : 2
## 1st Qu.: 7014
## Median : 26088
## Mean : 64210
## 3rd Qu.: 82706
## Max. :555609
## Audience <- Audience %>%
mutate(across(everything(),~ ifelse(is.na(.), mean(., na.rm=TRUE), .)))
summary(Audience)## original_title type genre runtime
## Length:651 Length:651 Length:651 Min. : 39.0
## Class :character Class :character Class :character 1st Qu.: 92.0
## Mode :character Mode :character Mode :character Median :103.0
## Mean :105.8
## 3rd Qu.:115.5
## Max. :267.0
## mpaa_rating imdb_rating imdb_num_votes critics_rating
## Length:651 Min. : 1.200 Min. : 180 Length:651
## Class :character 1st Qu.: 4.300 1st Qu.: 4035 Class :character
## Mode :character Median : 6.100 Median : 10651 Mode :character
## Mean : 5.811 Mean : 49378
## 3rd Qu.: 7.100 3rd Qu.: 56156
## Max. :55.000 Max. :893008
## critics_score audience_rating audience_score best_pic_nom
## Min. : 5.00 Length:651 Min. :13.00 Length:651
## 1st Qu.: 36.50 Class :character 1st Qu.:38.00 Class :character
## Median : 55.00 Mode :character Median :55.00 Mode :character
## Mean : 54.95 Mean :56.56
## 3rd Qu.: 73.50 3rd Qu.:76.00
## Max. :100.00 Max. :96.00
## Facebook_Likes
## Min. : 2
## 1st Qu.: 7014
## Median : 26088
## Mean : 64210
## 3rd Qu.: 82706
## Max. :555609Audience<- Audience |>
rename(
title = 'original_title')
Financials<-read.csv("Movie Dataset_Financials.csv")
View(Financials)
summary(Financials)## original_title budget..Millions. revenue..Millions. language
## Length:651 Min. : 0.32 Min. : 0.01 Length:651
## Class :character 1st Qu.: 10.00 1st Qu.: 20.75 Class :character
## Mode :character Median : 25.00 Median : 61.81 Mode :character
## Mean : 40.92 Mean : 137.97
## 3rd Qu.: 52.75 3rd Qu.: 154.49
## Max. :280.00 Max. :2068.18
## country
## Length:651
## Class :character
## Mode :character
##
##
## Financials<- Financials |>
rename(
title = 'original_title',
budget = 'budget..Millions.',
revenue = 'revenue..Millions.')
Audience<- Audience |>
mutate(
title = str_trim(title),
title = str_squish(title))
Financials<- Financials |>
mutate(
title = str_trim(title),
title = str_squish(title))
Audience |>
filter(duplicated(Audience$title))## title type genre runtime
## 1 The Legend of Tarzan Feature Film Drama 94
## 2 Pan Feature Film Other 90
## 3 The Twilight Saga: Breaking Dawn - Part 2 Feature Film Comedy 95
## 4 Fantastic Four Feature Film Drama 101
## 5 The Fast and the Furious Feature Film Drama 104
## 6 Godzilla Resurgence Feature Film Drama 98
## 7 Hercules Documentary Documentary 107
## mpaa_rating imdb_rating imdb_num_votes critics_rating critics_score
## 1 R 3.4 57933 Rotten 25
## 2 PG 3.6 1010 Fresh 33
## 3 PG 7.5 880 Fresh 90
## 4 R 2.1 9904 Fresh 35
## 5 R 7.1 128361 Fresh 68
## 6 PG-13 6.3 50340 Fresh 60
## 7 Unrated 3.8 10522 Fresh 21
## audience_rating audience_score best_pic_nom Facebook_Likes
## 1 Spilled 39 no 11175
## 2 Upright 24 no 393
## 3 Upright 89 no 359177
## 4 Upright 26 no 51261
## 5 Upright 78 no 125327
## 6 Spilled 67 no 699
## 7 Upright 20 no 235Audience |>
filter(title == "The Legend of Tarzan")## title type genre runtime mpaa_rating imdb_rating
## 1 The Legend of Tarzan Documentary Documentary 92 Unrated 6.8
## 2 The Legend of Tarzan Feature Film Drama 94 R 3.4
## imdb_num_votes critics_rating critics_score audience_rating audience_score
## 1 1942 Certified Fresh 66 Upright 68
## 2 57933 Rotten 25 Spilled 39
## best_pic_nom Facebook_Likes
## 1 no 121175
## 2 no 11175Audience<- Audience |>
filter(!duplicated(Audience$title))
Financials |>
filter(duplicated(Financials$title))## title budget revenue language country
## 1 The Legend of Tarzan 6.0 11.12 English USA
## 2 Pan 3.0 0.70 English USA
## 3 The Twilight Saga: Breaking Dawn - Part 2 0.5 1025.47 English USA
## 4 Fantastic Four 13.0 45.30 English USA
## 5 The Fast and the Furious 35.0 77.48 English USA
## 6 Godzilla Resurgence 1.9 157.11 Japanese Japan
## 7 Hercules 6.0 1.27 English USAFinancials |>
filter(title == "The Legend of Tarzan")## title budget revenue language country
## 1 The Legend of Tarzan 35 82.35 English USA
## 2 The Legend of Tarzan 6 11.12 English USAFinancials<- Financials |>
filter(!duplicated(Financials$title))
Movie_Data<-merge(Audience, Financials, by = "title")
View(Movie_Data)
summary(Movie_Data)## title type genre runtime
## Length:644 Length:644 Length:644 Min. : 39.0
## Class :character Class :character Class :character 1st Qu.: 92.0
## Mode :character Mode :character Mode :character Median :103.0
## Mean :105.9
## 3rd Qu.:116.0
## Max. :267.0
## mpaa_rating imdb_rating imdb_num_votes critics_rating
## Length:644 Min. : 1.200 Min. : 180 Length:644
## Class :character 1st Qu.: 4.300 1st Qu.: 4054 Class :character
## Mode :character Median : 6.150 Median : 10718 Mode :character
## Mean : 5.822 Mean : 49513
## 3rd Qu.: 7.100 3rd Qu.: 56142
## Max. :55.000 Max. :893008
## critics_score audience_rating audience_score best_pic_nom
## Min. : 5.00 Length:644 Min. :13.00 Length:644
## 1st Qu.: 37.00 Class :character 1st Qu.:38.00 Class :character
## Median : 55.00 Mode :character Median :55.50 Mode :character
## Mean : 55.03 Mean :56.64
## 3rd Qu.: 74.00 3rd Qu.:76.00
## Max. :100.00 Max. :96.00
## Facebook_Likes budget revenue language
## Min. : 2 Min. : 0.32 Min. : 0.01 Length:644
## 1st Qu.: 7309 1st Qu.: 10.00 1st Qu.: 21.39 Class :character
## Median : 26124 Median : 25.00 Median : 61.95 Mode :character
## Mean : 64057 Mean : 41.27 Mean : 137.43
## 3rd Qu.: 82664 3rd Qu.: 53.00 3rd Qu.: 154.25
## Max. :555609 Max. :280.00 Max. :2068.18
## country
## Length:644
## Class :character
## Mode :character
##
##
## What genres generate more revenue? What genres do critics have a preference for? Do critic scores or viewer scores matter more?
Creating a Correlogram Between all Variables
corrgram(Movie_Data, order=NULL,
panel=panel.shade,
text.panel=panel.txt,
col = colorRampPalette(c("white","red3")),
main="Correlogram")
Creating a Customer Segmentation Distribution
Making the bins for mpaa_rating more concise by renaming “Unrated” to “G”, and “NC-17” to “R”.
Movie_Data<- Movie_Data %>%
mutate(mpaa_rating = ifelse(mpaa_rating== "Unrated", "G", mpaa_rating))
View(Movie_Data)
Movie_Data<- Movie_Data %>%
mutate(mpaa_rating = ifelse(mpaa_rating== "NC-17", "R", mpaa_rating))
View(Movie_Data)Converting MPAA ratings to a factor to ensure proper ordering
Movie_Data$mpaa_rating <- factor(Movie_Data$mpaa_rating,
levels = c("G", "PG", "PG-13", "R"))Creating a bar chart of MPAA rating distribution
ggplot(Movie_Data, aes(x = mpaa_rating, fill = mpaa_rating)) +
geom_bar() +
labs(title = "Movie Distribution by Customer Segmentation",
x = "MPAA Rating",
y = "Count of Movies") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("G" = "goldenrod1", "PG" = "lightgoldenrod", "PG-13" = "red3", "R" = "salmon1"))
Creating a Genre Distribution
Shortening the genre titles
Movie_Data <- Movie_Data %>%
mutate(genre = ifelse(genre == "Action & Adventure", "Action", genre))
Movie_Data <- Movie_Data %>%
mutate(genre = ifelse(genre == "Art House & International", "Art House", genre))
Movie_Data <- Movie_Data %>%
mutate(genre = ifelse(genre == "Musical & Performing Arts", "Musical", genre))
Movie_Data <- Movie_Data %>%
mutate(genre = ifelse(genre == "Mystery & Suspense", "Mystery", genre))
Movie_Data <- Movie_Data %>%
mutate(genre = ifelse(genre == "Science Fiction & Fantasy", "Sci-Fi & Fantasy", genre))Creating a bar chart of genre distribution
ggplot(Movie_Data, aes(x = genre, fill = genre)) +
geom_bar() +
labs(title = "Movie Distribution by Genre",
x = "Genre",
y = "Count of Movies") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("Action" = "steelblue1", "Animation" = "royalblue1", "Art House" = "blue2", "Comedy" = "slateblue2", "Documentary" = "orchid3", "Drama" = "hotpink2", "Horror" = "brown2", "Musical" = "red3", "Mystery" = "salmon1", "Other" = "peachpuff", "Sci-Fi & Fantasy" = "navajowhite3"))
Combining the customer segmentation and genre
ggplot(Movie_Data, aes(x = genre, fill = mpaa_rating)) +
geom_bar(position = "stack") +
labs(title = "MPAA Rating Distribution Across Genres",
x = "Genre",
y = "Count of Movies") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("G" = "goldenrod1", "PG" = "lightgoldenrod",
"PG-13" = "red3", "R" = "salmon1"))
Revenue per Genre Graph
Summarizing the total revenue per genre
Genre_Revenue <- Movie_Data %>%
group_by(genre) %>%
summarise(total_revenue = sum(revenue, na.rm = TRUE)) %>%
arrange(desc(total_revenue))Creating the bar chart
ggplot(Genre_Revenue, aes(x = reorder(genre, -total_revenue), y = total_revenue, fill = genre)) +
geom_bar(stat = "identity") +
labs(title = "Total Revenue by Genre",
x = "Genre",
y = "Total Revenue (Millions)") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("Action" = "steelblue1", "Animation" = "royalblue1", "Art House" = "blue2",
"Comedy" = "slateblue2", "Documentary" = "orchid3", "Drama" = "hotpink2",
"Horror" = "brown2", "Musical" = "red3", "Mystery" = "salmon1",
"Other" = "peachpuff", "Sci-Fi & Fantasy" = "navajowhite3"))
Comparing Revenue to Budget and then Profit
Creating a profit column
Movie_Data <- Movie_Data %>%
mutate(profit = revenue - budget)Summarizing the revenue, budget, and profit per genre
Genre_Financials <- Movie_Data %>%
group_by(genre) %>%
summarise(
total_revenue = sum(revenue, na.rm = TRUE),
total_budget = sum(budget, na.rm = TRUE),
total_profit = sum(profit, na.rm = TRUE)
) %>%
pivot_longer(cols = c(total_revenue, total_budget, total_profit),
names_to = "Financial_Metric",
values_to = "Amount")Creating the grouped bar chart
ggplot(Genre_Financials, aes(x = genre, y = Amount, fill = Financial_Metric)) +
geom_bar(stat = "identity", position = "dodge") +
labs(title = "Comparison of Revenue, Budget, and Profit by Genre",
x = "Genre",
y = "Amount (Millions)") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("total_revenue" = "steelblue",
"total_budget" = "red3",
"total_profit" = "goldenrod1"))
Comparing critics’ and audience’s scores per genre
Summarizing the scores per genre
Genre_Scores <- Movie_Data %>%
group_by(genre) %>%
summarise(
avg_critics_score = mean(critics_score, na.rm = TRUE),
avg_audience_score = mean(audience_score, na.rm = TRUE)
) %>%
pivot_longer(cols = c(avg_critics_score, avg_audience_score),
names_to = "Score_Type",
values_to = "Average_Score")Creating a grouped bar chart
ggplot(Genre_Scores, aes(x = genre, y = Average_Score, fill = Score_Type)) +
geom_bar(stat = "identity", position = "dodge") +
labs(title = "Comparison of Critics' Scores and Audience Scores by Genre",
x = "Genre",
y = "Average Score") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("avg_critics_score" = "goldenrod1",
"avg_audience_score" = "red3"))
##### Simmarizing the average scores per genre and calculating the
overall average
Genre_Scores <- Movie_Data %>%
group_by(genre) %>%
summarise(
avg_critics_score = mean(critics_score, na.rm = TRUE),
avg_audience_score = mean(audience_score, na.rm = TRUE)
) %>%
mutate(avg_total_score = (avg_critics_score + avg_audience_score) / 2) %>%
pivot_longer(cols = c(avg_critics_score, avg_audience_score),
names_to = "Score_Type",
values_to = "Average_Score")Creating a grouped bar chart
ggplot(Genre_Scores, aes(x = genre)) +
geom_bar(aes(y = Average_Score, fill = Score_Type), stat = "identity", position = "dodge") +
geom_point(aes(y = avg_total_score), color = "black", size = 2) + # Smaller dots
geom_text(aes(y = avg_total_score + 5, label = round(avg_total_score, 1)), color = "black", size = 4) + # Labels above each genre
labs(title = "Comparison of Critics' Scores and Audience Scores by Genre",
x = "Genre",
y = "Average Score") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
scale_fill_manual(values = c("avg_critics_score" = "goldenrod1",
"avg_audience_score" = "red3"))
Predictive analysis: How are different factors driving overall revenue? Do these driving factors differ across genres?
Testing the correlation of the separate variables
Ensuring that all categorical factors are factors
Movie_Data <- Movie_Data %>%
mutate(genre = as.factor(genre),
country = as.factor(country))Excluding “title” and “profit” columns as they do not have an effect on revenue (on the other hand, revenue has an effect on profit, and title is irrelevant)
predictors <- setdiff(names(Movie_Data), c("title", "profit", "revenue"))Initializing an empty list to store models, then looping through each predictor and running a regression model
models <- list()
for (var in predictors) {
formula <- as.formula(paste("revenue ~", var))
model <- lm(formula, data = Movie_Data)
models[[var]] <- summary(model)
cat("\n\n### Regression Analysis for:", var, "###\n")
print(models[[var]])
}##
##
## ### Regression Analysis for: type ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -209.85 -114.96 -73.93 17.29 1930.37
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 123.93 28.36 4.369 1.45e-05 ***
## typeFeature Film 13.88 29.64 0.468 0.640
## typeTV Movie 114.13 97.43 1.171 0.242
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 208.4 on 641 degrees of freedom
## Multiple R-squared: 0.00217, Adjusted R-squared: -0.0009437
## F-statistic: 0.6969 on 2 and 641 DF, p-value: 0.4985
##
##
##
## ### Regression Analysis for: genre ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -207.01 -114.68 -71.36 16.52 1928.60
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 129.1185 25.9265 4.980 8.2e-07 ***
## genreAnimation 107.4338 74.3427 1.445 0.149
## genreArt House -2.3792 61.5876 -0.039 0.969
## genreComedy -0.8756 34.3545 -0.025 0.980
## genreDocumentary -9.4294 38.8898 -0.242 0.808
## genreDrama 10.4665 28.5891 0.366 0.714
## genreHorror 59.6289 50.7132 1.176 0.240
## genreMusical 8.2890 65.6748 0.126 0.900
## genreMystery 3.4965 37.5862 0.093 0.926
## genreOther 50.8565 61.5876 0.826 0.409
## genreSci-Fi & Fantasy -61.9151 74.3427 -0.833 0.405
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 209 on 633 degrees of freedom
## Multiple R-squared: 0.009001, Adjusted R-squared: -0.006654
## F-statistic: 0.5749 on 10 and 633 DF, p-value: 0.835
##
##
##
## ### Regression Analysis for: runtime ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -144.86 -117.09 -75.09 19.05 1922.14
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 109.8575 45.3640 2.422 0.0157 *
## runtime 0.2603 0.4213 0.618 0.5368
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 208.4 on 642 degrees of freedom
## Multiple R-squared: 0.0005944, Adjusted R-squared: -0.0009623
## F-statistic: 0.3819 on 1 and 642 DF, p-value: 0.5368
##
##
##
## ### Regression Analysis for: mpaa_rating ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -178.99 -112.26 -72.14 21.36 1888.89
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 143.04 25.19 5.678 2.07e-08 ***
## mpaa_ratingPG 36.25 31.73 1.142 0.254
## mpaa_ratingPG-13 -27.06 31.01 -0.872 0.383
## mpaa_ratingR -12.96 27.68 -0.468 0.640
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 207.8 on 640 degrees of freedom
## Multiple R-squared: 0.01017, Adjusted R-squared: 0.00553
## F-statistic: 2.192 on 3 and 640 DF, p-value: 0.08781
##
##
##
## ### Regression Analysis for: imdb_rating ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -807.01 -95.35 -66.52 4.50 1909.42
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 47.30 17.43 2.713 0.00684 **
## imdb_rating 15.48 2.66 5.819 9.32e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 203.2 on 642 degrees of freedom
## Multiple R-squared: 0.05011, Adjusted R-squared: 0.04863
## F-statistic: 33.87 on 1 and 642 DF, p-value: 9.325e-09
##
##
##
## ### Regression Analysis for: imdb_num_votes ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -163.42 -117.07 -74.35 17.45 1931.72
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.341e+02 9.194e+00 14.59 <2e-16 ***
## imdb_num_votes 6.679e-05 8.351e-05 0.80 0.424
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 208.4 on 642 degrees of freedom
## Multiple R-squared: 0.0009953, Adjusted R-squared: -0.0005608
## F-statistic: 0.6396 on 1 and 642 DF, p-value: 0.4241
##
##
##
## ### Regression Analysis for: critics_rating ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -174.23 -113.08 -77.42 18.45 1893.38
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 174.80 19.30 9.056 <2e-16 ***
## critics_ratingFresh -49.44 23.08 -2.142 0.0325 *
## critics_ratingRotten -41.56 23.24 -1.788 0.0742 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 207.9 on 641 degrees of freedom
## Multiple R-squared: 0.007376, Adjusted R-squared: 0.004279
## F-statistic: 2.382 on 2 and 641 DF, p-value: 0.09321
##
##
##
## ### Regression Analysis for: critics_score ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -282.11 -92.93 -45.45 31.74 1842.92
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -56.165 19.522 -2.877 0.00415 **
## critics_score 3.518 0.327 10.756 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 191.9 on 642 degrees of freedom
## Multiple R-squared: 0.1527, Adjusted R-squared: 0.1514
## F-statistic: 115.7 on 1 and 642 DF, p-value: < 2.2e-16
##
##
##
## ### Regression Analysis for: audience_rating ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -147.90 -116.95 -72.55 14.33 1936.29
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 148.09 14.05 10.543 <2e-16 ***
## audience_ratingUpright -16.20 17.31 -0.936 0.35
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 208.4 on 642 degrees of freedom
## Multiple R-squared: 0.001363, Adjusted R-squared: -0.0001929
## F-statistic: 0.876 on 1 and 642 DF, p-value: 0.3497
##
##
##
## ### Regression Analysis for: audience_score ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -276.38 -93.77 -27.03 25.10 1845.56
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -111.8470 21.2757 -5.257 2e-07 ***
## audience_score 4.4009 0.3524 12.489 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 187 on 642 degrees of freedom
## Multiple R-squared: 0.1955, Adjusted R-squared: 0.1942
## F-statistic: 156 on 1 and 642 DF, p-value: < 2.2e-16
##
##
##
## ### Regression Analysis for: best_pic_nom ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -137.85 -116.29 -75.38 16.70 1930.32
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 137.859 8.359 16.49 <2e-16 ***
## best_pic_nomyes -12.650 45.228 -0.28 0.78
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 208.5 on 642 degrees of freedom
## Multiple R-squared: 0.0001218, Adjusted R-squared: -0.001436
## F-statistic: 0.07823 on 1 and 642 DF, p-value: 0.7798
##
##
##
## ### Regression Analysis for: Facebook_Likes ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -434.73 -70.63 -42.44 54.57 1774.47
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.995e+01 7.267e+00 5.497 5.58e-08 ***
## Facebook_Likes 1.522e-03 6.411e-05 23.736 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 152.2 on 642 degrees of freedom
## Multiple R-squared: 0.4674, Adjusted R-squared: 0.4666
## F-statistic: 563.4 on 1 and 642 DF, p-value: < 2.2e-16
##
##
##
## ### Regression Analysis for: budget ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -410.22 -57.02 -22.27 18.26 1443.17
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 10.6674 8.0886 1.319 0.188
## budget 3.0717 0.1308 23.489 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 152.9 on 642 degrees of freedom
## Multiple R-squared: 0.4622, Adjusted R-squared: 0.4614
## F-statistic: 551.7 on 1 and 642 DF, p-value: < 2.2e-16
##
##
##
## ### Regression Analysis for: language ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -528.90 -113.12 -72.57 16.93 1933.67
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 18.66 205.19 0.091 0.92757
## languageEnglish 115.85 205.35 0.564 0.57283
## languageFrench 618.57 229.40 2.696 0.00719 **
## languageJapanese -14.50 290.18 -0.050 0.96016
## languageMandarin 81.87 290.18 0.282 0.77793
## languageSpanish 188.76 251.30 0.751 0.45286
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 205.2 on 638 degrees of freedom
## Multiple R-squared: 0.03754, Adjusted R-squared: 0.03
## F-statistic: 4.977 on 5 and 638 DF, p-value: 0.0001765
##
##
##
## ### Regression Analysis for: country ###
##
## Call:
## lm(formula = formula, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -275.29 -112.87 -70.26 18.33 1925.77
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 18.660 207.749 0.090 0.928
## countryAustralia 6.808 212.032 0.032 0.974
## countryBelgium 40.320 293.801 0.137 0.891
## countryCanada 84.850 213.145 0.398 0.691
## countryChina 44.055 254.439 0.173 0.863
## countryFrance 260.622 217.888 1.196 0.232
## countryGermany 105.081 216.986 0.484 0.628
## countryJapan 204.990 254.439 0.806 0.421
## countryMexico 153.180 293.801 0.521 0.602
## countryNew Line 84.380 293.801 0.287 0.774
## countryNew Zealand 304.753 239.887 1.270 0.204
## countrySpain 66.840 293.801 0.228 0.820
## countryUK 93.249 210.614 0.443 0.658
## countryUSA 123.748 207.944 0.595 0.552
##
## Residual standard error: 207.7 on 630 degrees of freedom
## Multiple R-squared: 0.02572, Adjusted R-squared: 0.005614
## F-statistic: 1.279 on 13 and 630 DF, p-value: 0.2203Regression analysis for top correlations (critics_score, audience_score, budget, Facebook_Likes)
Defining the multiple regression model
multi_reg_model <- lm(revenue ~ critics_score + audience_score + budget + Facebook_Likes, data = Movie_Data)
summary(multi_reg_model)##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = Movie_Data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -302.80 -54.62 -5.03 29.55 1505.24
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -8.792e+01 1.445e+01 -6.084 2.02e-09 ***
## critics_score 7.803e-01 3.369e-01 2.316 0.0209 *
## audience_score 7.919e-01 3.847e-01 2.058 0.0400 *
## budget 1.838e+00 1.282e-01 14.339 < 2e-16 ***
## Facebook_Likes 9.631e-04 6.163e-05 15.628 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 125.1 on 639 degrees of freedom
## Multiple R-squared: 0.6415, Adjusted R-squared: 0.6392
## F-statistic: 285.8 on 4 and 639 DF, p-value: < 2.2e-16Extracting R-squared and adjusted R-squared, computing VIF (Variance Inflation Factor) to check multicollinearity, and calculating R
r_squared <- summary(multi_reg_model)$r.squared
adj_r_squared <- summary(multi_reg_model)$adj.r.squared
vif_values <- vif(multi_reg_model)
r_value <- sqrt(r_squared)
cat("\n### Multiple Regression Analysis ###\n")##
## ### Multiple Regression Analysis ###cat("R (Multiple Correlation Coefficient):", round(r_value, 4), "\n")## R (Multiple Correlation Coefficient): 0.8009cat("R-Squared:", round(r_squared, 4), "\n")## R-Squared: 0.6415cat("Adjusted R-Squared:", round(adj_r_squared, 4), "\n")## Adjusted R-Squared: 0.6392cat("\n### Variance Inflation Factors (VIF) ###\n")##
## ### Variance Inflation Factors (VIF) ###print(vif_values)## critics_score audience_score budget Facebook_Likes
## 2.496051 2.662210 1.434645 1.366296Regression Scatter Plot
Predicting revenue using the model
Movie_Data$predicted_revenue <- predict(multi_reg_model, newdata = Movie_Data)Creating the scatter plot
ggplot(Movie_Data, aes(x = predicted_revenue, y = revenue)) +
geom_point(color = "red3") + # Red dots for actual revenue
geom_smooth(method = "lm", color = "black", se = FALSE) + # Black regression line
labs(title = "Actual vs Predicted Revenue",
x = "Predicted Revenue",
y = "Actual Revenue") +
theme_minimal()## `geom_smooth()` using formula = 'y ~ x'
Residual Plot
Checking to make sure the predictive analysis does not have any errors
Movie_Data$residuals <- residuals(multi_reg_model)
ggplot(Movie_Data, aes(x = predicted_revenue, y = residuals)) +
geom_point(color = "red3") + # Red dots for residuals
geom_hline(yintercept = 0, linetype = "solid", color = "black") + # Zero line
labs(title = "Residual Plot",
x = "Predicted Revenue",
y = "Residuals") +
theme_minimal()
Seeing if these driving factors differ across genres
Initializing an empty list to store models, then looping through each predictor (critics_score, audience_score, budget and Facebook_Likes) and running a regression model
genre_models <- list()
for (g in unique(Movie_Data$genre)) {
genre_data <- Movie_Data %>% filter(genre == g)
if (nrow(genre_data) > 5) {
model <- lm(revenue ~ critics_score + audience_score + budget + Facebook_Likes, data = genre_data)
genre_models[[g]] <- model
cat("\n### Regression Analysis for Genre:", g, "###\n")
print(summary(model))
}
}##
## ### Regression Analysis for Genre: Drama ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -304.64 -57.11 -7.19 32.02 1506.05
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -9.219e+01 2.554e+01 -3.610 0.00036 ***
## critics_score 9.414e-01 5.751e-01 1.637 0.10272
## audience_score 7.351e-01 6.335e-01 1.160 0.24680
## budget 1.803e+00 2.104e-01 8.567 5.93e-16 ***
## Facebook_Likes 9.751e-04 1.048e-04 9.308 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 143.4 on 296 degrees of freedom
## Multiple R-squared: 0.5723, Adjusted R-squared: 0.5666
## F-statistic: 99.03 on 4 and 296 DF, p-value: < 2.2e-16
##
##
## ### Regression Analysis for Genre: Mystery ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -201.302 -54.812 4.024 46.818 260.410
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -1.196e+02 3.229e+01 -3.705 0.000499 ***
## critics_score 1.215e+00 7.945e-01 1.529 0.132041
## audience_score 1.183e+00 9.659e-01 1.225 0.226019
## budget 8.550e-01 2.872e-01 2.977 0.004346 **
## Facebook_Likes 1.224e-03 1.295e-04 9.451 4.86e-13 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 83.88 on 54 degrees of freedom
## Multiple R-squared: 0.8193, Adjusted R-squared: 0.8059
## F-statistic: 61.21 on 4 and 54 DF, p-value: < 2.2e-16
##
##
## ### Regression Analysis for Genre: Comedy ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -222.725 -29.912 6.187 31.010 304.637
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -8.828e+01 2.322e+01 -3.802 0.000277 ***
## critics_score 7.361e-01 5.453e-01 1.350 0.180806
## audience_score 7.733e-01 6.429e-01 1.203 0.232484
## budget 2.717e+00 2.521e-01 10.776 < 2e-16 ***
## Facebook_Likes 6.500e-04 1.199e-04 5.421 5.96e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 75.9 on 81 degrees of freedom
## Multiple R-squared: 0.8527, Adjusted R-squared: 0.8454
## F-statistic: 117.2 on 4 and 81 DF, p-value: < 2.2e-16
##
##
## ### Regression Analysis for Genre: Action ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -183.63 -45.21 -10.25 24.78 398.10
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -1.025e+02 3.673e+01 -2.789 0.007069 **
## critics_score 7.980e-01 6.473e-01 1.233 0.222507
## audience_score 1.543e+00 8.198e-01 1.882 0.064666 .
## budget 1.785e+00 2.765e-01 6.453 2.12e-08 ***
## Facebook_Likes 5.825e-04 1.582e-04 3.681 0.000499 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 92.47 on 60 degrees of freedom
## Multiple R-squared: 0.6928, Adjusted R-squared: 0.6723
## F-statistic: 33.82 on 4 and 60 DF, p-value: 9.172e-15
##
##
## ### Regression Analysis for Genre: Musical ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -85.54 -30.96 -10.47 17.33 103.04
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.068e+01 7.050e+01 -0.435 0.67656
## critics_score 1.672e+00 1.807e+00 0.925 0.38556
## audience_score -1.418e+00 2.366e+00 -0.599 0.56779
## budget 1.198e+00 1.076e+00 1.114 0.30214
## Facebook_Likes 9.781e-04 2.380e-04 4.109 0.00452 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 70.65 on 7 degrees of freedom
## Multiple R-squared: 0.8775, Adjusted R-squared: 0.8075
## F-statistic: 12.54 on 4 and 7 DF, p-value: 0.00262
##
##
## ### Regression Analysis for Genre: Documentary ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -235.25 -54.36 -8.73 24.57 667.37
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -6.363e+01 5.256e+01 -1.211 0.2321
## critics_score -2.029e+00 2.146e+00 -0.945 0.3493
## audience_score 3.447e+00 2.177e+00 1.584 0.1200
## budget 1.136e+00 5.093e-01 2.230 0.0305 *
## Facebook_Likes 1.181e-03 2.237e-04 5.277 3.28e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 126.6 on 47 degrees of freedom
## Multiple R-squared: 0.603, Adjusted R-squared: 0.5692
## F-statistic: 17.84 on 4 and 47 DF, p-value: 5.669e-09
##
##
## ### Regression Analysis for Genre: Other ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -122.83 -27.15 10.52 28.90 140.88
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -9.549e+01 7.246e+01 -1.318 0.2201
## critics_score 3.445e+00 3.382e+00 1.019 0.3350
## audience_score -2.168e+00 3.297e+00 -0.657 0.5273
## budget 1.875e+00 7.250e-01 2.586 0.0294 *
## Facebook_Likes 9.942e-04 3.964e-04 2.508 0.0334 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 87.4 on 9 degrees of freedom
## Multiple R-squared: 0.9243, Adjusted R-squared: 0.8906
## F-statistic: 27.47 on 4 and 9 DF, p-value: 4.663e-05
##
##
## ### Regression Analysis for Genre: Sci-Fi & Fantasy ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## 1 2 3 4 5 6 7 8 9
## -11.257 -48.980 21.789 37.591 -22.903 8.863 57.863 -39.486 -3.481
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 27.289215 107.553515 0.254 0.812
## critics_score -4.485788 4.217079 -1.064 0.347
## audience_score 5.826905 3.451658 1.688 0.167
## budget 0.434039 0.600002 0.723 0.509
## Facebook_Likes -0.001476 0.002227 -0.663 0.544
##
## Residual standard error: 49.84 on 4 degrees of freedom
## Multiple R-squared: 0.8073, Adjusted R-squared: 0.6146
## F-statistic: 4.189 on 4 and 4 DF, p-value: 0.0971
##
##
## ### Regression Analysis for Genre: Horror ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -370.63 -63.16 13.64 41.01 419.68
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 5.279e+00 1.005e+02 0.053 0.959
## critics_score 3.561e+00 2.540e+00 1.402 0.178
## audience_score -4.835e+00 3.265e+00 -1.481 0.156
## budget 6.410e+00 1.116e+00 5.745 1.91e-05 ***
## Facebook_Likes -3.066e-04 6.155e-04 -0.498 0.624
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 154.5 on 18 degrees of freedom
## Multiple R-squared: 0.8303, Adjusted R-squared: 0.7926
## F-statistic: 22.02 on 4 and 18 DF, p-value: 9.872e-07
##
##
## ### Regression Analysis for Genre: Art House ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -107.88 -24.02 -19.11 -10.46 201.82
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 11.0671704 86.0991271 0.129 0.9005
## critics_score -0.2932747 1.3387724 -0.219 0.8315
## audience_score 0.4643615 1.9218791 0.242 0.8145
## budget 1.7541955 0.7315859 2.398 0.0400 *
## Facebook_Likes 0.0006518 0.0003148 2.070 0.0683 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 88.56 on 9 degrees of freedom
## Multiple R-squared: 0.5835, Adjusted R-squared: 0.3983
## F-statistic: 3.152 on 4 and 9 DF, p-value: 0.07044
##
##
## ### Regression Analysis for Genre: Animation ###
##
## Call:
## lm(formula = revenue ~ critics_score + audience_score + budget +
## Facebook_Likes, data = genre_data)
##
## Residuals:
## 1 2 3 4 5 6 7 8 9
## 29.48 18.48 49.67 20.80 -75.24 30.96 -43.04 26.81 -57.93
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.338e+02 7.156e+01 -3.267 0.03087 *
## critics_score 2.370e+00 2.697e+00 0.879 0.42921
## audience_score 4.257e+00 1.828e+00 2.329 0.08031 .
## budget -1.740e+00 1.105e+00 -1.575 0.19049
## Facebook_Likes 1.483e-03 1.475e-04 10.057 0.00055 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 64.53 on 4 degrees of freedom
## Multiple R-squared: 0.9739, Adjusted R-squared: 0.9479
## F-statistic: 37.36 on 4 and 4 DF, p-value: 0.002003Identifying if the driving factors of revenue differ for each genre by extracting the top 2 driving factors per genre
genre_factors <- data.frame(Genre = character(), Factor = character(), Coefficient = numeric(), p_value = numeric())
for (g in names(genre_models)) {
model <- genre_models[[g]]
if (!inherits(model, "lm")) {
next
}
coef_summary <- summary(model)$coefficients
if (nrow(coef_summary) > 1) {
coefficients <- coef_summary[-1, 1]
p_values <- coef_summary[-1, 4]
df <- data.frame(
Genre = g,
Factor = rownames(coef_summary)[-1],
Coefficient = as.numeric(coefficients),
p_value = as.numeric(p_values)
)
df <- df %>% filter(p_value < 0.05) %>% arrange(desc(abs(Coefficient))) %>% head(2)
if (nrow(df) > 0) {
genre_factors <- rbind(genre_factors, df)
}
}
}
print(genre_factors)## Genre Factor Coefficient p_value
## 1 Drama budget 1.8026849718 5.932222e-16
## 2 Drama Facebook_Likes 0.0009750676 3.041403e-18
## 3 Mystery budget 0.8550309531 4.345513e-03
## 4 Mystery Facebook_Likes 0.0012238953 4.857759e-13
## 5 Comedy budget 2.7165068523 2.594605e-17
## 6 Comedy Facebook_Likes 0.0006499796 5.955437e-07
## 7 Action budget 1.7845083534 2.123393e-08
## 8 Action Facebook_Likes 0.0005824697 4.988575e-04
## 9 Musical Facebook_Likes 0.0009781188 4.520852e-03
## 10 Documentary budget 1.1359108139 3.053536e-02
## 11 Documentary Facebook_Likes 0.0011807621 3.276095e-06
## 12 Other budget 1.8747488437 2.942088e-02
## 13 Other Facebook_Likes 0.0009942402 3.340487e-02
## 14 Horror budget 6.4103721361 1.908042e-05
## 15 Art House budget 1.7541954945 4.004217e-02
## 16 Animation Facebook_Likes 0.0014831791 5.497405e-04Visualizing the top 2 driving factors in each genre
for (g in unique(genre_factors$Genre)) {
genre_data <- Movie_Data %>% filter(genre == g)
factors <- genre_factors %>% filter(Genre == g) %>% pull(Factor)
if (length(factors) >= 2) {
factor1 <- factors[1]
factor2 <- factors[2]
p1 <- ggplot(genre_data, aes_string(x = factor1, y = "revenue")) +
geom_point(color = "red3") +
geom_smooth(method = "lm", color = "black", se = FALSE) +
labs(title = paste("Revenue vs", factor1, "in", g), x = factor1, y = "Revenue") +
theme_minimal()
p2 <- ggplot(genre_data, aes_string(x = factor2, y = "revenue")) +
geom_point(color = "blue3") +
geom_smooth(method = "lm", color = "black", se = FALSE) +
labs(title = paste("Revenue vs", factor2, "in", g), x = factor2, y = "Revenue") +
theme_minimal()
print(p1)
print(p2)
}
}## Warning: `aes_string()` was deprecated in ggplot2 3.0.0.
## ℹ Please use tidy evaluation idioms with `aes()`.
## ℹ See also `vignette("ggplot2-in-packages")` for more information.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
Overall recommendations for the company
Looking at the ideal move based on 100% controllable variables (genre, runtime, budget, language, and country)
Processing data: converting categorical values to factors, one-hot encoding categorical variables, normalizing continuous variables, and removing original categorical columns
Movie_Data$genre <- as.factor(Movie_Data$genre)
Movie_Data$language <- as.factor(Movie_Data$language)
Movie_Data$country <- as.factor(Movie_Data$country)
dummies <- dummyVars(~ genre + language + country, data = Movie_Data)
Movie_Data_encoded <- predict(dummies, newdata = Movie_Data) %>%
as.data.frame()
Movie_Data_encoded$budget <- Movie_Data$budget
Movie_Data_encoded$runtime <- Movie_Data$runtime
Movie_Data_encoded$revenue <- Movie_Data$revenue
Movie_Data_encoded$budget <- scale(Movie_Data_encoded$budget)
Movie_Data_encoded$runtime <- scale(Movie_Data_encoded$runtime)
str(Movie_Data_encoded)## 'data.frame': 644 obs. of 34 variables:
## $ genre.Action : num 0 0 0 0 1 0 0 1 0 1 ...
## $ genre.Animation : num 0 0 0 0 0 0 0 0 0 0 ...
## $ genre.Art House : num 0 0 0 0 0 0 0 0 0 0 ...
## $ genre.Comedy : num 0 0 0 1 0 0 1 0 0 0 ...
## $ genre.Documentary : num 0 0 0 0 0 0 0 0 0 0 ...
## $ genre.Drama : num 1 1 0 0 0 1 0 0 1 0 ...
## $ genre.Horror : num 0 0 0 0 0 0 0 0 0 0 ...
## $ genre.Musical : num 0 0 0 0 0 0 0 0 0 0 ...
## $ genre.Mystery : num 0 0 1 0 0 0 0 0 0 0 ...
## $ genre.Other : num 0 0 0 0 0 0 0 0 0 0 ...
## $ genre.Sci-Fi & Fantasy: num 0 0 0 0 0 0 0 0 0 0 ...
## $ language. : num 1 0 0 0 0 0 0 0 0 0 ...
## $ language.English : num 0 1 1 1 1 1 1 1 1 1 ...
## $ language.French : num 0 0 0 0 0 0 0 0 0 0 ...
## $ language.Japanese : num 0 0 0 0 0 0 0 0 0 0 ...
## $ language.Mandarin : num 0 0 0 0 0 0 0 0 0 0 ...
## $ language.Spanish : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country. : num 1 0 0 0 0 0 0 0 0 0 ...
## $ country.Australia : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.Belgium : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.Canada : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.China : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.France : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.Germany : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.Japan : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.Mexico : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.New Line : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.New Zealand : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.Spain : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.UK : num 0 0 0 0 0 0 0 0 0 0 ...
## $ country.USA : num 0 1 1 1 1 1 1 1 1 1 ...
## $ budget : num [1:644, 1] -0.622 0.081 -0.548 -0.819 -0.548 ...
## ..- attr(*, "scaled:center")= num 41.3
## ..- attr(*, "scaled:scale")= num 46.1
## $ runtime : num [1:644, 1] 1.44 0.108 -0.456 -0.405 0.261 ...
## ..- attr(*, "scaled:center")= num 106
## ..- attr(*, "scaled:scale")= num 19.5
## $ revenue : num 18.66 60.22 31.56 0.75 19.68 ...Training a predictive model, by first ensuring reproducibility, then doing a Random Forest model (combining the output of multiple decision trees to reach a single result), and checking the variable importance
colSums(is.na(Movie_Data_encoded))## genre.Action genre.Animation genre.Art House
## 0 0 0
## genre.Comedy genre.Documentary genre.Drama
## 0 0 0
## genre.Horror genre.Musical genre.Mystery
## 0 0 0
## genre.Other genre.Sci-Fi & Fantasy language.
## 0 0 0
## language.English language.French language.Japanese
## 0 0 0
## language.Mandarin language.Spanish country.
## 0 0 0
## country.Australia country.Belgium country.Canada
## 0 0 0
## country.China country.France country.Germany
## 0 0 0
## country.Japan country.Mexico country.New Line
## 0 0 0
## country.New Zealand country.Spain country.UK
## 0 0 0
## country.USA
## 0 0 0
## revenue
## 0Movie_Data_encoded <- na.omit(Movie_Data_encoded)
colnames(Movie_Data_encoded) <- make.names(colnames(Movie_Data_encoded))
colnames(Movie_Data_encoded)## [1] "genre.Action" "genre.Animation" "genre.Art.House"
## [4] "genre.Comedy" "genre.Documentary" "genre.Drama"
## [7] "genre.Horror" "genre.Musical" "genre.Mystery"
## [10] "genre.Other" "genre.Sci.Fi...Fantasy" "language."
## [13] "language.English" "language.French" "language.Japanese"
## [16] "language.Mandarin" "language.Spanish" "country."
## [19] "country.Australia" "country.Belgium" "country.Canada"
## [22] "country.China" "country.France" "country.Germany"
## [25] "country.Japan" "country.Mexico" "country.New.Line"
## [28] "country.New.Zealand" "country.Spain" "country.UK"
## [31] "country.USA" "budget" "runtime"
## [34] "revenue"set.seed(123)
model <- randomForest(revenue ~ ., data = Movie_Data_encoded, ntree = 500, importance = TRUE)
importance(model)## %IncMSE IncNodePurity
## genre.Action -2.0629982 1.285042e+05
## genre.Animation 0.8712342 2.167023e+05
## genre.Art.House 1.8466578 5.991725e+04
## genre.Comedy 1.4165968 2.097757e+05
## genre.Documentary -3.9351418 2.195749e+05
## genre.Drama 4.1107679 3.759883e+05
## genre.Horror 2.2360427 2.345085e+05
## genre.Musical -2.7764470 8.044547e+04
## genre.Mystery -3.4439895 2.052570e+05
## genre.Other 0.2049438 1.498764e+05
## genre.Sci.Fi...Fantasy 1.4957504 5.558148e+04
## language. 0.0000000 1.679182e+02
## language.English -2.2226092 2.344806e+05
## language.French 1.6651317 4.675654e+05
## language.Japanese 0.0000000 1.485003e+03
## language.Mandarin 0.0000000 1.360147e+03
## language.Spanish 1.2965258 4.560342e+03
## country. 0.0000000 4.672061e+02
## country.Australia 5.9409368 1.139381e+05
## country.Belgium 0.0000000 2.097353e+03
## country.Canada -2.3234044 6.494410e+04
## country.China -1.2078405 1.494641e+03
## country.France -2.4240310 1.463674e+05
## country.Germany 0.7570995 3.222123e+04
## country.Japan -2.0851743 4.062564e+04
## country.Mexico 0.0000000 1.057997e+03
## country.New.Line 0.0000000 9.782750e+03
## country.New.Zealand 2.1551053 1.255662e+05
## country.Spain 0.0000000 1.258903e+03
## country.UK -4.4383242 1.209512e+05
## country.USA 2.7834660 2.397310e+05
## budget 61.5999583 1.465527e+07
## runtime -1.0363890 3.131929e+06varImpPlot(model)
##### Finding the best movie configuration and creating the ideal movie,
then predicting revenue for this movie
print(colnames(Movie_Data_encoded))## [1] "genre.Action" "genre.Animation" "genre.Art.House"
## [4] "genre.Comedy" "genre.Documentary" "genre.Drama"
## [7] "genre.Horror" "genre.Musical" "genre.Mystery"
## [10] "genre.Other" "genre.Sci.Fi...Fantasy" "language."
## [13] "language.English" "language.French" "language.Japanese"
## [16] "language.Mandarin" "language.Spanish" "country."
## [19] "country.Australia" "country.Belgium" "country.Canada"
## [22] "country.China" "country.France" "country.Germany"
## [25] "country.Japan" "country.Mexico" "country.New.Line"
## [28] "country.New.Zealand" "country.Spain" "country.UK"
## [31] "country.USA" "budget" "runtime"
## [34] "revenue"best_movie <- as.data.frame(t(rep(0, length(colnames(Movie_Data_encoded)))))
colnames(best_movie) <- colnames(Movie_Data_encoded)
best_movie$runtime <- scale(120, center = attr(scale(Movie_Data_encoded$runtime), "scaled:center"),
scale = attr(scale(Movie_Data_encoded$runtime), "scaled:scale"))
best_movie$budget <- scale(200000000, center = attr(scale(Movie_Data_encoded$budget), "scaled:center"),
scale = attr(scale(Movie_Data_encoded$budget), "scaled:scale"))
best_movie$genre.Action...Adventure <- 1 # Set exact name found in Step 1
best_movie$language.English <- 1
best_movie$country.USA <- 1
predicted_revenue <- predict(model, newdata = best_movie)
print(predicted_revenue)## 1
## 1008.441Looking at the predicted revenue compared to dataset
revenue_df <- data.frame(revenue = Movie_Data_encoded$revenue)
# Create a density plot
ggplot(revenue_df, aes(x = revenue)) +
geom_density(fill = "blue3", alpha = 0.3) + # Density curve of actual revenues
geom_vline(aes(xintercept = predicted_revenue), color = "red3", linetype = "dashed", size = 1) + # Mark predicted revenue
labs(title = "Predicted Revenue vs. Movie Revenue Distribution",
x = "Revenue",
y = "Density") +
theme_minimal()## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
What are the predicted box office sales for a movie screening with a 55 critic score and 1,250 FaceBook likes? What is the effect of a marketing campaign?
Predicted box office sales (before campaign)
intercept <- -87.92
critics_coef <- 0.7803
facebook_coef <- 0.0009631
critics_score <- 55
facebook_likes_initial <- 1250
predicted_revenue_initial <- intercept + (critics_coef * critics_score) +
(facebook_coef * facebook_likes_initial)
cat("Predicted Revenue for Premiere: $", round(predicted_revenue_initial, 2), "\n")## Predicted Revenue for Premiere: $ -43.8Predicted box office sales (after campaign)
facebook_likes_new <- facebook_likes_initial * 100 # Increase by 10,000%
predicted_revenue_new <- intercept + (critics_coef * critics_score) +
(facebook_coef * facebook_likes_new)
cat("Predicted Revenue After Advertising Campaign: $", round(predicted_revenue_new, 2), "\n")## Predicted Revenue After Advertising Campaign: $ 75.38Calculating the excpected increase
increase <- predicted_revenue_new - predicted_revenue_initial
cat("Expected Revenue Increase: $", round(increase, 2), "\n")## Expected Revenue Increase: $ 119.18Creating a bar chart
revenue_data <- data.frame(
Scenario = c("Before Campaign", "After Campaign"),
Revenue = c(predicted_revenue_initial, predicted_revenue_new)
)
revenue_data$Scenario <- factor(revenue_data$Scenario, levels = c("Before Campaign", "After Campaign"))
ggplot(revenue_data, aes(x = Scenario, y = Revenue, fill = Scenario)) +
geom_bar(stat = "identity", width = 0.6, show.legend = FALSE) +
scale_fill_manual(values = c("goldenrod1", "red3")) + # Custom colors
geom_text(aes(label = round(Revenue, 2)), vjust = -0.5, size = 5) +
labs(title = "Predicted Revenue Before and After Advertising Campaign",
x = "Scenario",
y = "Predicted Revenue ($)") +
theme_minimal()