library(dplyr)
set.seed(123)
transactions <- data.frame(
 transaction_id = 1:1000,
 user_id = sample(1:50, 1000, replace = TRUE),
 amount = round(runif(1000, 1, 200), 2),
 status = sample(c("success", "fail"), 1000, replace = TRUE, prob = c(0.8, 0.2)),
 region_id = sample(1:5, 1000, replace = TRUE)
)

users <- data.frame(
 user_id = 1:50,
 region_id = sample(1:5, 50, replace = TRUE)
)

# Основний аналіз
result <- transactions %>%
 # Фільтруємо тільки успішні транзакції
 filter(status == "success") %>%
 group_by(user_id) %>%
 summarise(total_spend = sum(amount)) %>%
 left_join(users, by = "user_id") %>%

 group_by(region_id) %>%
 mutate(avg_region_spend = mean(total_spend)) %>%
 ungroup() %>%

 filter(total_spend > avg_region_spend) %>%

 arrange(region_id, desc(total_spend))

print(result)

library(ggplot2)

ggplot(result, aes(x = factor(region_id), y = total_spend)) +
 geom_boxplot(aes(y = avg_region_spend), fill = "lightblue", alpha = 0.5) +
 geom_point(aes(y = total_spend), color = "red", size = 3) +
 labs(title = "Користувачі з витратами вище середнього по регіонах",
      x = "Регіон",
      y = "Сумарні витрати (TC)") +
 theme_minimal()