# the last line is projected based on citations to the end of october
# (almost certainly conservative)
library(ggplot2)

d <- read.csv("wikipedia_citations.txt",header=F)
colnames(d) <- c("year", "citations")

# print the total number of citations
sum(d$citations)

# generate and print a graph
# p <- qplot(year, citations, data=d) +
#  geom_line(colour="blue") + geom_point(colour="blue")

p <- qplot(factor(year), citations, data=d, geom="bar", fill=I("darkblue"))
p <- p + scale_x_discrete("Year") + scale_y_continuous("Number of Papers")

pdf("citations_by_year.pdf", width=7.5, height=5.3)
print(p)
dev.off()

## data from dario
##########################################################3

# import data from dario
d <- read.csv("Wikipedia publications - Data.csv")

# clean up the dates
colnames(d)[1] <- "date"
d <- d[,c(-9,-10)]
d <- d[!d$date == "2013 to date",]
d$date <- as.factor(d$date)

library(reshape)
qplot(date, value, data=melt(d), group=variable, geom="line") +
  aes(colour=variable) + scale_y_log10()