setwd('C:\\Users\\Student\\Downloads\\Sztuczna Inteligencja')
movie_body_counts <- read.csv("filmdeathcounts.csv", stringsAsFactors=FALSE)

library (dplyr)
library (ggplot2)

head(movie_body_counts)
str(movie_body_counts)

movie_body_counts$body_per_min <- movie_body_counts$Body_Cou nt/movie_body_counts$Length_Minutes
ggplot(movie_body_counts, aes(x=Body_Count)) + geom_histogram(bins=20, color="grey", fill="lightblue")

movie_body_counts %>%
  top_n(n = 10, Body_Count) %>%
  arrange(desc(Body_Count))

movie_body_counts %>%
  top_n(n = 10, Length_Minutes) %>%
  arrange(desc(Length_Minutes))

ggplot(movie_body_counts, aes(x=IMDB_Rating)) + geom_histogram(bins=10, color="grey", fill="lightblue")
#new

set.seed(900)
imdb_simulation <- rnorm(n=nrow(movie_body_counts), mean=imdb_mean, sd=imdb_mean)
movie_body_counts$imdb_simulation <- imdb_simulation
ggplot(movie_body_counts, aes(x=imdb_simulation)) + geom_histogram(bins=10, color="grey", fill="lightblue")
ggplot(movie_body_counts, aes(sample=imdb_simulation)) + stat_qq()

ggplot(movie_body_counts, aes(sample=IMDB_Rating)) + stat_qq()