Week 3 - Snapchat Assignment

Question 1: Read in Snapchat Report

snap <- read_csv("Snap_Transparency_Report_H1-2024.csv")

## Rows: 22988 Columns: 7
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (7): period, section, category, sub_category_1, sub_category_2, metric, ...
## 
## 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.

Class of the value column:

class(snap$value)

## [1] "character"

is.numeric(snap$value)    # FALSE

## [1] FALSE

is.character(snap$value)  # TRUE

## [1] TRUE

Convert to numeric:

snap$value <- as.numeric(snap$value)
class(snap$value)

## [1] "numeric"

Question 2: Wide vs Long Data?

This data would be considered Long Data since the rows outweight the number of coloumns present. There are 22988 observations yet only 7 variables

Question 3: Subset to Snapchat Enforcement

snap_enforcement <- snap %>%
  filter(section == "Overview of Our T&S Enforcements",
         category == "Country")

Question 4: Remove Repetitive Columns and Rename Country

snap_enforcement <- snap_enforcement %>%
  select(-period, -section, -category)

snap_enforcement <- snap_enforcement %>%
  rename(country = sub_category_1)

Question 5: Pivot Country and Sub Category

snap_enforcement <- snap_enforcement %>%
  unite("country_category", country, sub_category_2, sep = "-", remove = FALSE) %>%
  pivot_wider(names_from = metric, values_from = value)

head(snap_enforcement)

## # A tibble: 6 × 6
##   country_category                     country sub_category_2 total_enforcements
##   <chr>                                <chr>   <chr>                       <dbl>
## 1 Afghanistan-Child Sexual Exploitati… Afghan… Child Sexual …                340
## 2 Afghanistan-Drugs                    Afghan… Drugs                          27
## 3 Afghanistan-False Information        Afghan… False Informa…                  3
## 4 Afghanistan-Harassment and Bullying  Afghan… Harassment an…                726
## 5 Afghanistan-Hate Speech              Afghan… Hate Speech                     5
## 6 Afghanistan-Impersonation            Afghan… Impersonation                   4
## # ℹ 2 more variables: total_unique_accounts_enforced <dbl>,
## #   median_turnaround_time_minutes <dbl>

Question 6: Percentage of Enforcements from the United States

US_Enforcements <- snap_enforcement %>%
  group_by(country) %>%
  summarise(total_enforcements = sum(total_enforcements)) %>%
  mutate(percentage = 100 * total_enforcements / sum(total_enforcements)) %>%
  arrange(desc(percentage)) %>%
  filter(country == "United States")

US_Enforcements

## # A tibble: 1 × 3
##   country       total_enforcements percentage
##   <chr>                      <dbl>      <dbl>
## 1 United States            7389396       38.1

Question 7: Worldwide Weapons Enforcements

Weapon_Enforcements <- snap_enforcement %>%
  filter(sub_category_2 == "Weapons") %>%
  summarise("Worldwide Weapon Enforcements" = sum(total_enforcements))

Weapon_Enforcements

## # A tibble: 1 × 1
##   `Worldwide Weapon Enforcements`
##                             <dbl>
## 1                          234512

Question 8: Enforcements per Unique Account

snap_enforcement <- snap_enforcement %>%
  mutate(enforcements_per_account = total_enforcements / total_unique_accounts_enforced)

Highest_enforced <- snap_enforcement %>%
  arrange(desc(enforcements_per_account)) %>%
  select(country, sub_category_2, total_enforcements,
         total_unique_accounts_enforced, enforcements_per_account)

head(Highest_enforced, 5)

## # A tibble: 5 × 5
##   country          sub_category_2      total_enforcements total_unique_account…¹
##   <chr>            <chr>                            <dbl>                  <dbl>
## 1 Paraguay         Weapons                             14                      1
## 2 Sierra Leone     Child Sexual Explo…                121                     13
## 3 South Korea      Child Sexual Explo…               4211                    526
## 4 Marshall Islands Sexual Content                      11                      2
## 5 Vietnam          Weapons                              5                      1
## # ℹ abbreviated name: ¹​total_unique_accounts_enforced
## # ℹ 1 more variable: enforcements_per_account <dbl>

Question 9: EU Weapons Enforcement vs. Response Time

data("eu_countries")

eu_names <- eu_countries$name
eu_names[eu_names == "Czechia"] <- "Czech Republic"

eu_weapons <- snap_enforcement %>%
  filter(sub_category_2 == "Weapons", country %in% eu_names)

ggplot(eu_weapons, aes(x = enforcements_per_account,
                       y = median_turnaround_time_minutes,
                       color = total_enforcements,
                       label = country)) +
  geom_point(size = 3, alpha = 1) +
  geom_text_repel(size = 3, color = "black", max.overlaps = Inf) +
  scale_y_log10() +
  scale_color_viridis_c(trans = "log10") +
  labs(x = "Enforcements per unique account",
       y = "Median enforcement time (minutes)",
       color = "Total enforcements",
       title = "Snapchat Weapons T&S response: enforcement intensity vs. time")

Interpretation: Countries with more enforcements per account tend to have a faster response time. France, Belgium, the Netherlands and Luxembourg are in the fast, and France also has by far the highest volume.