Goal: Predict the average movie rating

Import data

horror_movies <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2022/2022-11-01/horror_movies.csv')
## Rows: 32540 Columns: 20
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr  (10): original_title, title, original_language, overview, tagline, post...
## dbl   (8): id, popularity, vote_count, vote_average, budget, revenue, runtim...
## lgl   (1): adult
## date  (1): release_date
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
skimr::skim(horror_movies)
Data summary
Name horror_movies
Number of rows 32540
Number of columns 20
_______________________
Column type frequency:
character 10
Date 1
logical 1
numeric 8
________________________
Group variables None

Variable type: character

skim_variable n_missing complete_rate min max empty n_unique whitespace
original_title 0 1.00 1 191 0 30296 0
title 0 1.00 1 191 0 29563 0
original_language 0 1.00 2 2 0 97 0
overview 1286 0.96 1 1000 0 31020 0
tagline 19835 0.39 1 237 0 12513 0
poster_path 4474 0.86 30 32 0 28048 0
status 0 1.00 7 15 0 4 0
backdrop_path 18995 0.42 29 32 0 13536 0
genre_names 0 1.00 6 144 0 772 0
collection_name 30234 0.07 4 56 0 815 0

Variable type: Date

skim_variable n_missing complete_rate min max median n_unique
release_date 0 1 1950-01-01 2022-12-31 2012-12-09 10999

Variable type: logical

skim_variable n_missing complete_rate mean count
adult 0 1 0 FAL: 32540

Variable type: numeric

skim_variable n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist
id 0 1.00 445910.83 305744.67 17 146494.8 426521.00 707534.00 1033095.00 ▇▆▆▅▅
popularity 0 1.00 4.01 37.51 0 0.6 0.84 2.24 5088.58 ▇▁▁▁▁
vote_count 0 1.00 62.69 420.89 0 0.0 2.00 11.00 16900.00 ▇▁▁▁▁
vote_average 0 1.00 3.34 2.88 0 0.0 4.00 5.70 10.00 ▇▂▆▃▁
budget 0 1.00 543126.59 4542667.81 0 0.0 0.00 0.00 200000000.00 ▇▁▁▁▁
revenue 0 1.00 1349746.73 14430479.15 0 0.0 0.00 0.00 701842551.00 ▇▁▁▁▁
runtime 0 1.00 62.14 41.00 0 14.0 80.00 91.00 683.00 ▇▁▁▁▁
collection 30234 0.07 481534.88 324498.16 656 155421.0 471259.00 759067.25 1033032.00 ▇▅▅▅▅ 
data <- horror_movies %>%
    
    # Treat missing values
    select(-tagline, -backdrop_path, -collection_name, -collection) %>% 
    na.omit() %>% 
    
    # Log transform variables with pos-skewed distribution
    mutate(popularity = log(popularity))

Explore data

Identify good predictors

Popularity

data %>% 
    ggplot(aes(vote_average, popularity)) +
    scale_x_log10() +
    geom_point()
## Warning: Transformation introduced infinite values in continuous x-axis

Revenue

data %>% 
    ggplot(aes(vote_average, revenue)) +
    geom_point()

Runtime

data %>% 
    ggplot(aes(vote_average, runtime)) +
    geom_point()

Title

data %>%
    
    # Tokenize title
    unnest_tokens(output = word, input = title) %>%
    
    # Calculate avg rating per word
    group_by(word) %>%
    summarise(vote_average = mean(vote_average),
              n = n()) %>%
    ungroup() %>%
    
    filter(n > 5, !str_detect(word, "\\d")) %>%
    slice_max(order_by = vote_average, n = 25) %>%
    
    # Plot
    ggplot(aes(vote_average, fct_reorder(word, vote_average))) +
    geom_point() +
    
    labs(y = "Words in Title")