Wrangling dataset by picking out Rotten Tomatoes movie ratings from a column - r

Question

I have this sample dataset:
structure(list(Title = c("Isn't It Romantic", "Isn't It Romantic",
"Isn't It Romantic", "Isn't It Romantic", "Isn't It Romantic",
"Isn't It Romantic", "Gully Boy", "Gully Boy", "Gully Boy", "Gully Boy",
"Gully Boy", "Gully Boy", "The Wandering Earth", "The Wandering Earth",
"The Wandering Earth", "The Wandering Earth", "The Wandering Earth",
"The Wandering Earth", "How to Train Your Dragon: The Hidden World",
"How to Train Your Dragon: The Hidden World", "How to Train Your Dragon: The Hidden World",
"How to Train Your Dragon: The Hidden World", "How to Train Your Dragon: The Hidden World",
"How to Train Your Dragon: The Hidden World", "American Woman",
"American Woman", "Us", "Us", "Us", "Us", "Us", "Us", "The Wolf's Call",
"The Wolf's Call", "Avengers: Endgame", "Avengers: Endgame",
"Avengers: Endgame", "Avengers: Endgame", "Avengers: Endgame",
"Avengers: Endgame", "The Silence", "The Silence", "The Silence",
"The Silence", "The Silence", "The Silence", "My Little Pony: Equestria Girls: Spring Breakdown",
"My Little Pony: Equestria Girls: Spring Breakdown"), Ratings = c("Internet Movie Database",
"5.9/10", "Rotten Tomatoes", "68%", "Metacritic", "60/100", "Internet Movie Database",
"8.4/10", "Rotten Tomatoes", "100%", "Metacritic", "65/100",
"Internet Movie Database", "6.4/10", "Rotten Tomatoes", "74%",
"Metacritic", "62/100", "Internet Movie Database", "7.6/10",
"Rotten Tomatoes", "91%", "Metacritic", "71/100", "Rotten Tomatoes",
"57%", "Internet Movie Database", "7.1/10", "Rotten Tomatoes",
"94%", "Metacritic", "81/100", "Internet Movie Database", "7.6/10",
"Internet Movie Database", "8.7/10", "Rotten Tomatoes", "94%",
"Metacritic", "78/100", "Internet Movie Database", "5.2/10",
"Rotten Tomatoes", "23%", "Metacritic", "25/100", "Internet Movie Database",
"7.7/10")), row.names = c(NA, -48L), class = c("tbl_df", "tbl",
"data.frame"))
The Ratings column has 3 different types of Ratings (Imdb, Rotten Tomatoes and Metacritic) for each movie, spread out over 6 rows for each movie.
I'd like to wrangle this dataset so that for each movie, I create a new column called rottentomatoes_rating and the values are the rating. So, in my sample dataset, Isn't it Romantic movie would have 68% under rottentomatoes_rating, Gully Boy would have 100% under rottentomatoes_rating, etc.
For those movies that don't have a rottentomatoes_rating, then I'd like to put NA under rottentomatoes_rating.
I've thought about using spread in tidyr, but I can't quite figure out how to do so since in my case, the variable and values are all in the same column!

Answer 1

Assuming your dataset is called dt you can use this process to get a tidy version of your dataset:
library(tidyverse)
# specify indexes of Rating companies
ids = seq(1, nrow(dt), 2)
# get rows of Rating companies
dt %>% slice(ids) %>%
# combine with the rating values
cbind(dt %>% slice(-ids) %>% select(RatingsValue = Ratings)) %>%
# reshape dataset
spread(Ratings, RatingsValue)
# Title Year Rated Released Runtime Internet Movie Database Metacritic Rotten Tomatoes
# 1 Gully Boy 2019 Not Rated 2019-02-14 153 min 8.4/10 65/100 100%
# 2 Isn't It Romantic 2019 PG-13 2019-02-13 89 min 5.9/10 60/100 68%

Answer 2

If the data is formatted similarly throughout your dataset, the following code should work:
df %>% group_by(Title) %>%
slice(match("Rotten Tomatoes", df$Ratings) + 1) %>%
rename(rottentomatoes_rating = Ratings)
This gives:
# A tibble: 2 x 6
# Groups: Title [2]
Title Year Rated Released Runtime rottentomatoes_rating
<chr> <chr> <chr> <date> <chr> <chr>
1 Gully Boy 2019 Not Rated 2019-02-14 153 min 100%
2 Isn't It Romantic 2019 PG-13 2019-02-13 89 min 68%
For the NAs, if the original data always has the RT score the row after the string is observed, then it should give you NA by default.

Answer 3

sumshyftw answer is good.
But here is a data.table version if you simply want to get rotten tomatoes's percents:
dt <- dt[dt$Ratings %like% "%",]
dt <- setnames(dt, "Ratings", "rottentomatoes_rating")
Output :
# A tibble: 2 x 6
Title Year Rated Released Runtime rottentomatoes_rating
<chr> <chr> <chr> <date> <chr> <chr>
1 Isn't It Romantic 2019 PG-13 2019-02-13 89 min 68%
2 Gully Boy 2019 Not Rated 2019-02-14 153 min 100%
I used %like% "%" because I assume that the full data is just like your example.

Answer 4

new version that fills NA values when blank
# using data.table
library(data.table)
dt <- as.data.table(df)
# Index will hold whether the row is a Provider eg Rotten Tomatoes, or a value
dt[, Index:=rep(c("Provider", "Value"), .N/2)]
# Need an index to bind these together
dt[, Provider.Id:=rep(1:(.N/2), each=2), by=Title]
dt[1:6,]
# segment out the Provider & Values in to columns
out <- dcast(dt, Title+Provider.Id~Index, value.var = "Ratings")
out[, Provider := NULL]
# now convert to full wide format
out_df <- as.data.frame(dcast(out, Title~Provider, value.var="Value", fill=NA))
out_df

Answer 5

To get all metrics with data.table
# using data.table
library(data.table)
dt <- as.data.table(df)
# groups the data set with by, and extracts the Ratings
# makes use of logic that the odd indeces hold the name of the provider,
# the even ones hold the values. Only works if this holds.
# It can probably be optimised a bit. dcast converts from long to required wide
# format
splitRatings <- function(Ratings){
# e.g. Ratings=dt$Ratings[1:6]
N <- length(Ratings)
split_dt <- data.table(DB=Ratings[1:N %% 2 == 1],
Values=Ratings[1-(1:N %% 2) == 1])
out <- dcast(split_dt, .~DB, value.var = "Values")
out[, ".":=NULL]
out
}
# applies the function based on the by clause, returning the table embedded
dt2 <- dt[, splitRatings(Ratings), by=.(Title, Year, Rated, Released, Runtime)]
# convert back
out <- as.data.frame(dt2)

Answer 6

Here is one version.
df %>%
mutate(Value = ifelse(str_detect(Ratings, "\\d"), Ratings, NA)) %>%
fill(Value, .direction = "up") %>%
filter(!str_detect(Ratings, "\\d")) %>%
spread(Ratings, Value)