---
title: "Game of Thrones"
author: "Rachel Bellflowers"
output: 
  flexdashboard::flex_dashboard:
    favicon: favicon.png
    theme: yeti
    orientation: rows
    vertical_layout: scroll
    source_code: embed
    social: menu
---

```{r setup, include = FALSE}
knitr::opts_chunk$set(comment = "")
library(flexdashboard)
library(tidyverse)
library(xml2)
library(rvest)
library(data.table)
library(knitr)
library(dygraphs)
library(plotly)
library(DT)
```

```{r webscrape, include = FALSE}
link_01 <- "https://www.imdb.com/title/tt0944947/episodes?season=1"
link_02 <- "https://www.imdb.com/title/tt0944947/episodes?season=2"
link_03 <- "https://www.imdb.com/title/tt0944947/episodes?season=3"
link_04 <- "https://www.imdb.com/title/tt0944947/episodes?season=4"
link_05 <- "https://www.imdb.com/title/tt0944947/episodes?season=5"
link_06 <- "https://www.imdb.com/title/tt0944947/episodes?season=6"
link_07 <- "https://www.imdb.com/title/tt0944947/episodes?season=7"
link_08 <- "https://www.imdb.com/title/tt0944947/episodes?season=8"

# Season 1

seas_1_episodes <- read_html(link_01) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_1_ratings <- read_html(link_01) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_1_dates <- read_html(link_01) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()

# Season 2

seas_2_episodes <- read_html(link_02) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_2_ratings <- read_html(link_02) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_2_dates <- read_html(link_02) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()


# Season 3

seas_3_episodes <- read_html(link_03) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_3_ratings <- read_html(link_03) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_3_dates <- read_html(link_03) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()

# Season 4

seas_4_episodes <- read_html(link_04) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_4_ratings <- read_html(link_04) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_4_dates <- read_html(link_04) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()

# Season 5

seas_5_episodes <- read_html(link_05) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_5_ratings <- read_html(link_05) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_5_dates <- read_html(link_05) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()

# Season 6

seas_6_episodes <- read_html(link_06) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_6_ratings <- read_html(link_06) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_6_dates <- read_html(link_06) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()

# Season 7

seas_7_episodes <- read_html(link_07) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_7_ratings <- read_html(link_07) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_7_dates <- read_html(link_07) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()

# Season 8

seas_8_episodes <- read_html(link_08) %>%
  html_nodes(xpath = "//*[(@id = \"episodes_content\")]//strong//a") %>%
  html_text()

seas_8_ratings <- read_html(link_08) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star\", \" \" )) and contains(concat( \" \", @class, \" \" ), concat( \" \", \"small\", \" \" ))]//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"ipl-rating-star__rating\", \" \" ))]") %>%
  html_text()

seas_8_dates <- read_html(link_08) %>%
  html_nodes(xpath = "//*[contains(concat( \" \", @class, \" \" ), concat( \" \", \"airdate\", \" \" ))]") %>%
  html_text()
```


```{r manipulate, include = FALSE}
# Seasons

hmm <- replicate(n = 10, "Season 1", simplify = TRUE)
hmm2 <- replicate(n = 10, "Season 2", simplify = TRUE)
hmm3 <- replicate(n = 10, "Season 3", simplify = TRUE)
hmm4 <- replicate(n = 10, "Season 4", simplify = TRUE)
hmm5 <- replicate(n = 10, "Season 5", simplify = TRUE)
hmm6 <- replicate(n = 10, "Season 6", simplify = TRUE)
hmm7 <- replicate(n = 7, "Season 7", simplify = TRUE)
hmm8 <- replicate(n = 6, "Season 8", simplify = TRUE)

# Episode number

hmms <- replicate(n = 6, c(1:10), simplify = TRUE)
episode_num <- c(hmms, 1:7, 1:6)

# Combine into datatable

dates <- c(seas_1_dates, seas_2_dates, seas_3_dates, seas_4_dates, seas_5_dates, seas_6_dates, seas_7_dates, seas_8_dates)

episodes <- c(seas_1_episodes, seas_2_episodes, seas_3_episodes, seas_4_episodes, seas_5_episodes, seas_6_episodes, seas_7_episodes, seas_8_episodes)

ratings <- c(seas_1_ratings, seas_2_ratings, seas_3_ratings, seas_4_ratings, seas_5_ratings, seas_6_ratings, seas_7_ratings, seas_8_ratings)

seasons <- c(hmm, hmm2, hmm3, hmm4, hmm5, hmm6, hmm7, hmm8)

got <- data.table(seasons, episodes, episode_num, ratings, dates)

got$dates <- str_trim(got$dates)

# Changing data types

got$ratings <- as.numeric(got$ratings)

got$dates <- str_remove_all(got$dates, "\\.")
got$dates <- as.Date(got$dates, "%d %b %Y")
got$year <- as.numeric(format(got$dates, "%Y"))
```


Episodes
=======================================================================

Row
-----------------------------------------------------------------------
### Data Table

```{r}
datatable(got)
```

Row
-----------------------------------------------------------------------
### Episode Ratings by Season

```{r bar_chart}
bar_chart <- ggplot(got, aes(factor(episode_num), ratings, fill = seasons)) +
  geom_col() +
  facet_grid(~seasons) +
  labs(x = "Episode Number", y = "Rating") +
  theme_minimal() 

bar_chart_p <- ggplotly(bar_chart)

bar_chart_p <- bar_chart_p %>% layout(showlegend = FALSE)

ggplotly(bar_chart_p)
```

Seasons
=======================================================================

Row
-----------------------------------------------------------------------

### Time Series of Show Ratings

```{r time}
# This is a time-series graph with the title: Time Series of Show Ratings. 
# It shows how ratings for the show changed throughout the years from 2011 to 2018. 
# Years appears on the x-axis. Ratings out of 10 appear on the y-axis. Tick marks for the y-axis are at 4, 5, 6, 7, 8, 9, and 10. 
#The values are as follows: 2011 = rating of 9.1; 2012 = rating of 8.8; 2013 = rating of 8.8; 2014 = rating of 9.1; 2015 = rating of 8.5; 2016 = rating of 8.5; 2017 = rating of 8.6; 2018 = rating of 7.5.

series <- got[ , c(6, 4)]

dygraph(series)
```

### Boxplot of Ratings

```{r boxplot}
plot_ly(got, y = ~ratings, type = "box")

# Less robust against outliers:

#mean(got$ratings) # 8.838356
 
#sd(got$ratings) # 0.9346167

# More robust against outliers:

#median(got$ratings) # 8.9

#IQR(got$ratings) # 0.7
```

Project Description
=======================================================================

This is a project I did on IMDB ratings of Game of Thrones. 

I web-scraped the data from IMDB using `rvest` and `xml2`, then I created this dashboard using `flexdashboard`, `plotly`, `ggplot2`, `dygraphs`, and `DT`.

To view the full repository and the license for this project, follow this [link](https://github.com/rachelbellflowers/gameofthrones).