Code to create graphics for Anki -> R article on http://chrisk91.me

create.R

require(RSQLite)
require(DBI)
require(ggplot2)
require(jsonlite)
require(ggthemes)

save_plot <- function(plot, filename, widthpx = 800, heightpx = 500){
  dpi <- 120
  width <- widthpx / dpi
  height <- heightpx / dpi
  ggsave(filename, plot, width = width, height = height, units = "in", dpi = dpi)
}

connection <- dbConnect(SQLite(), dbname="collection.anki2")

result <- dbSendQuery(connection, "SELECT * from cards")
cardData <- dbFetch(result)
dbClearResult(result)

result <- dbSendQuery(connection, "SELECT * from col")
collectionData <- dbFetch(result)
dbClearResult(result)

result <- dbSendQuery(connection, "SELECT * from notes")
notesData <- dbFetch(result)
dbClearResult(result)

result <- dbSendQuery(connection, "SELECT * from revlog")
revisionData <- dbFetch(result)
dbClearResult(result)

dbDisconnect(connection)


collectionNames <- fromJSON(txt=collectionData$decks[1])

getDeckName <- function(x) collectionNames[[x]]$name
cardData$deck <- factor(mapply(getDeckName, as.character(cardData$did))) #perform lookup of deck names


getDeckNameFromCid <- function(x){
    lookup <- as.character(cardData$deck[cardData$id == x])
    
    if(identical(lookup, character(0))) #due to deleted cards, some lookups might fail
    {
      lookup <- "Removed card"
    }
    
    return(lookup)
  }
revisionData$deck <- factor(mapply(getDeckNameFromCid, revisionData$cid))


tmp <- ggplot(revisionData, aes(time / 1000)) + geom_histogram(binwidth = 10) + xlab("time spent reviewing a single card\n(seconds)") + scale_y_log10() + ggtitle("Review times")
save_plot(tmp + theme_minimal(), "histogram.png")

tmp <- ggplot(revisionData, aes(time / 1000, fill = deck)) + guides(fill=FALSE) + geom_histogram(binwidth = 10) + xlim(0, 400) + xlab("time spent reviewing a single card\n(seconds)") + scale_y_log10() + ggtitle("Review times") + facet_wrap(~ deck)
save_plot(tmp + theme_minimal() + ggthemes::scale_fill_gdocs(), "histogram_decks.png", widthpx = 800, heightpx = 800)

tmp <- ggplot(cardData, aes(ivl, fill = deck)) + guides(fill=FALSE) + geom_histogram(binwidth = 10) + facet_wrap(~ deck, scales = "free_y") + xlab("interval")
save_plot(tmp + theme_minimal() + ggthemes::scale_fill_gdocs(), "histogram_review_times.png", widthpx = 800, heightpx = 800)


plotWidth <- floor(sqrt(nrow(cardData))) # our final dimesnion should be a square
cardData$index <- (0 : (nrow(cardData) - 1)) # to calculate the position of each card, we need a running count
cardData$X <- cardData$index %% plotWidth # x position
cardData$Y <- cardData$index %/% plotWidth # y position
cardData$mature <- cardData$ivl > 21 # you can also use this, to show mature cards
tmp <- ggplot(cardData, aes(X, Y)) + geom_tile(aes(fill = ivl)) + scale_fill_gradient2(mid = "#dd6a27", high = "#12e824")
save_plot(tmp + theme_void() + guides(fill = FALSE), "review_heatmap.png", widthpx = 400, heightpx = 400)

# Lets filter out premade decks
cardDataFiltered <- cardData[!(cardData$deck %in% c("Sprachen::Französisch::5000 Most Common French Words", "Sprachen::Französisch::Verben mit Konjugationen")), ]
plotWidth <- floor(sqrt(nrow(cardDataFiltered))) # our final dimesnion should be a square
cardDataFiltered$index <- (0 : (nrow(cardDataFiltered) - 1)) # to calculate the position of each card, we need a running count
cardDataFiltered$X <- cardDataFiltered$index %% plotWidth # x position
cardDataFiltered$Y <- cardDataFiltered$index %/% plotWidth # y position
cardDataFiltered$mature <- cardDataFiltered$ivl > 21 # you can also use this, to show mature cards
tmp <- ggplot(cardDataFiltered, aes(X, Y)) + geom_tile(aes(fill = mature))
save_plot(tmp + theme_void() + guides(fill = FALSE), "review_heatmap_filtered.png", widthpx = 400, heightpx = 400)