In stephan-koenig/nrdata:

knitr::opts_chunk$set(echo = TRUE)

This R Markdown document is made interactive using Shiny. Unlike the more traditional workflow of creating static reports, you can now create documents that allow your readers to change the assumptions underlying your analysis and see the results immediately.

To learn more, see Interactive Documents.

Inputs and Outputs

You can embed Shiny inputs and outputs in your document. Outputs are automatically updated whenever inputs change. This demonstrates how a standard R plot can be made interactive by wrapping it in the Shiny renderPlot function. The selectInput and sliderInput functions create the input widgets used to drive the plot.

inputPanel(
  selectInput("n_breaks", label = "Number of bins:",
              choices = c(10, 20, 35, 50), selected = 20),

  sliderInput("bw_adjust", label = "Bandwidth adjustment:",
              min = 0.2, max = 2, value = 1, step = 0.2)
)

renderPlot({
  hist(faithful$eruptions, probability = TRUE, breaks = as.numeric(input$n_breaks),
       xlab = "Duration (minutes)", main = "Geyser eruption duration")

  dens <- density(faithful$eruptions, adjust = input$bw_adjust)
  lines(dens, col = "blue")
})

Embedded Application

It's also possible to embed an entire Shiny application within an R Markdown document using the shinyAppDir function. This example embeds a Shiny application located in another directory:

shinyAppDir(
  system.file("examples/06_tabsets", package = "shiny"),
  options = list(
    width = "100%", height = 550
  )
)

Note the use of the height parameter to determine how much vertical space the embedded application should occupy.

You can also use the shinyApp function to define an application inline rather then in an external directory.

In all of R code chunks above the echo = FALSE attribute is used. This is to prevent the R code within the chunk from rendering in the document alongside the Shiny components.

library(dplyr)
library(ggplot2)

load("nrdata.RData")

ggplot(nrdata, aes(win_type, turns_played)) +
       geom_boxplot()

by(nrdata$turns_played, nrdata$win_type, summary)
summary(nrdata$runner_score[nrdata$win_type == "agenda" &
                            nrdata$winner == "runner"])
ggplot(nrdata[nrdata$win_type == "agenda" & nrdata$winner == "corp", ],
       aes(turns_played - 10)) +
    geom_histogram(binwidth = 1, boundary = 0)

plot_that_stuff <- function(pack, myfilename) {

    alpha_value <- 0.8

    pack_data <- filter(nrdata, release == pack)

    all_ids <- nrdata %>%
        group_by(corp_id, runner_id) %>% 
        summarize(total = n()) %>%
        select(corp_id, runner_id) %>% ungroup()

    total_games <- pack_data %>%
        group_by(corp_id, runner_id) %>% 
        summarize(total = n()) %>% ungroup() %>%
        arrange(corp_id, runner_id)

    corp_wins <- pack_data %>%
        filter(winner == "corp") %>%
        group_by(corp_id, runner_id) %>%
        summarize(corp_wins = n()) %>% ungroup()

    corp_wins <- full_join(corp_wins, all_ids, by = c("corp_id", "runner_id"))

    corp_wins[is.na(corp_wins$corp_wins), "corp_wins"] <- 0

    corp_wins <- ungroup(corp_wins) %>% arrange(corp_id, runner_id)

    all_games <- sum(total_games$total)

    combined <- full_join(total_games, corp_wins, by = c("corp_id", "runner_id"))

    combined <- mutate(combined, corp_percent = corp_wins/total * 100,
                       percent = total/all_games * 100)

    ggplot(combined, aes(runner_id, corp_id,
                         size = percent,
                         colour = corp_percent)) +
        geom_point(alpha = alpha_value, na.rm = TRUE) +
        scale_size_area(limits = c(0,10),
                        name = "Matchup freq [%]") +
        labs(title = pack,
             x = "Runner", y = "Corp") +
        scale_colour_gradient2(low = "Red", high = "Blue", mid = "grey97",
                              midpoint = 50, na.value = "grey95",
                              limits = c(0,100),
                              name = "Corp win [%]") +
        theme_bw() +
        theme(axis.text.x = element_text(angle = 45, hjust = 1),
              axis.ticks = element_blank(),
              panel.background = element_blank(),
              panel.border = element_blank(),
              panel.grid.major = element_line(colour = "grey95"),
              legend.key = element_rect(colour = NA)) +
              guides(colour = guide_legend(override.aes = list(alpha = alpha_value)))
        ggsave(paste0("~/Downloads/", myfilename),
               width=8, height=6)
}

# pack <- data_packs[14]
# mapply(plot_that_stuff, pack, "test.jpg")

data_packs <- sort(unique(nrdata$release))
filenames <- unlist(lapply(1:19, paste0, ".pdf"))

mapply(plot_that_stuff, data_packs, filenames)

The Shifting Meta

I wanted to practice data analysis in R and decided to take closer look at the last OCTGN data set that was made available in Feb 2015 thanks to db0.

Here the first preliminary graph that shows a global view of the meta over time, i.e. changes with each new release (from Trace Amount to The Source):

Each frame is a snap shot of the meta for a specific RELEASE (indicated in the title) of all players on OCTGN.
Each column is a single RUNNER.
Each row is a single CORP.
Each circle represents the specific matchup between the runner and the corp of the intersection (e.g. NEH vs. Andromeda).
The size of each circle indicates MATCHUP FREQUENCY during a specific release (i.e number of games of a specific matchup divided by total number of games).
The color of each circle represents the CORP WIN RATE of that matchup during a specific release, with light grey representing 50% (i.e split), shades of blue > 50% (i.e. favorably for corp), and shades of red a corp win rate < 50% (i.e. favorably for runner).

Here some general guidelines on how to interpret the graph:

If a lot of circles in a single row are large then the corresponding corp is popular.
If a lot of circles in a single row are blue then the corp is successful.
Correspondingly, large, red circles in a single column indicate a popular and succesful runner.
If a lot of circles in all rows and columns are blue then corps dominate the meta, if red then runners.

Some stray observations:

Newly released IDs are initally popular.
Notice how the "Stronger Together" row is mostly red? Always.
The meta drastically shifts blue when Honor and Profit is released.
After its release, NEH surges while MN and TWIY decline.

The next goals for further analysis are: Getting data for later releases. Implement interactive graphs (e.g. a slider to manually switch between releases, look only at the top 25% of players).

What trends and patterns do you see in the data? What parameters would you like to explore? Please let me know any feedback in the comments!

## To show groups with 0 members.
# df <- data.frame(
#   "ID"   = rep(1:2, each = 2),
#   "Col1" = c("A", NA, "AA", NA),
#   "Col2" = c("B", "C", "BB", "CC"))
# 
# df %>% 
#  group_by(ID) %>% 
#  do(complete(., Col1, Col2, fill = list(ID = .$ID)))