Merge pull request #23 from KonScience/develop

katrinleinweber · katrinleinweber · commit b93e8ddf568a · 2015-05-02T14:53:39.000+02:00
restyle code
diff --git a/summarize-flattr-reports.R b/summarize-flattr-reports.R
@@ -1,37 +1,31 @@
-# READ ME: https://github.com/KonScience/Summarize-Flattr-Reports#summarize-flattr-reports
+# Please read https://github.com/KonScience/Summarize-Flattr-Reports#summarize-flattr-reports
+
+rm(list = ls())  # clean workspace
+original_wd <- getwd()  # save current working directory
+Sys.setlocale("LC_ALL", "UTF-8")  # respect non-ASCII symbols like German umlauts on Mac OSX, learned from https://stackoverflow.com/questions/8145886/
+options(stringsAsFactors = FALSE, row.names = FALSE, limitsize = FALSE) # set global options
 
-# load packages for data frame manipulation & diagram drawing
 # see http://www.r-bloggers.com/library-vs-require-in-r/ for require() vs. library() discussion
 library(scales)
 library(ggplot2)
 library(plyr)
 
 # get all filenames of Flattr Monthly Revenue CSV; assumes that all were downloaded into same folder
-
 args <- commandArgs(trailingOnly = TRUE)
-
 if (length(args) == 0) { # execute via: Rscript path/to/summarize-flattr-reports.R path/to/flattr-revenue-000000.csv
   print("Please select one of the 'flattr-revenue-....csv' files from the folder you downloaded them to.")
   first_flattr_file <- file.choose()
   flattr_dir <- dirname(first_flattr_file) # learned from http://stackoverflow.com/a/18003224
 } else {
   if ((substring(args[1], 1, 1) == "/") || (substring(args[1], 2, 2) == ":")) {
     flattr_dir <- dirname(args[1]) # set absolute directory by cli argument
-  } else {
-    flattr_dir <- dirname(file.path(getwd(), args[1], fsep = .Platform$file.sep)) # set relative directory by cli argument
-  }
+  } else {flattr_dir <- dirname(file.path(getwd(), args[1], fsep = .Platform$file.sep))} # set relative directory by cli argument
 }
-
 Flattr_filenames <- list.files(flattr_dir, pattern = "flattr-revenue-20[0-9]{4}.csv")
-
-# move working directory to .csv files but save original
-original_wd <- getwd()
 setwd(flattr_dir)
-options(stringsAsFactors = FALSE)
 
-# check for summary file of previously processed data & add new reports, instead of reading in every files again
+# use summary file if available & create if not, instead of reading files individually
 try(known_raw <- read.csv2("flattr-revenue-000000.csv", encoding = "UTF-8"))
-
 if ("flattr-revenue-000000.csv" %in% list.files(flattr_dir, pattern = "*.csv")) {
   # check for existing raw date & merge with new
   if (length(unique(known_raw$period)) < length(Flattr_filenames)) {
@@ -55,20 +49,14 @@ if ("flattr-revenue-000000.csv" %in% list.files(flattr_dir, pattern = "*.csv"))
                           encoding = "UTF-8" # learned from RTFM, but works only on Win7
                    ))  # Function structure learned from https://stat.ethz.ch/pipermail/r-help/2010-October/255593.html
   }} else {raw <- do.call("rbind", lapply(Flattr_filenames, read.csv2, encoding = "UTF-8"))}  # same as inner else, just to catch edge case of repetive plotting without adding new Revenue Reports
-
-Sys.setlocale("LC_ALL", "UTF-8")  # respect non-ASCII symbols like German umlauts on Mac OSX, learned from https://stackoverflow.com/questions/8145886/
-
-# export aggregated data for next (month's) run
-write.csv2(raw, "flattr-revenue-000000.csv", row.names = FALSE)
+write.csv2(x = raw, file = "flattr-revenue-000000.csv")
 
 # append 1st days to months & convert to date format; learned from http://stackoverflow.com/a/4594269
 raw$period <- as.Date(paste(raw$period, "-01"), format="%Y-%m -%d")
 raw$EUR_per_click <- raw$revenue / raw$clicks
 
 # populate raw data with all_revenue for each thing
-for (i in 1:dim(raw)[1]){
-  raw$all_revenue[i] <- sum(subset(raw, title == raw$title[i])$revenue)
-}
+for (i in 1:nrow(raw)){raw$all_revenue[i] <- sum(subset(raw, title == raw$title[i])$revenue)}
 
 # determine dataset size to auto-adjust plots
 N_months <- length(Flattr_filenames)
@@ -81,21 +69,16 @@ per_thing <- ddply(.data = raw,
                    all_clicks = sum(clicks),
                    all_revenue = sum(revenue))
 per_thing <- per_thing[order(per_thing$all_revenue, decreasing = TRUE),]
-rownames(per_thing) <- NULL
-write.csv2(x = per_thing,
-           file = "flattr-revenue-things.csv",
-           row.names = FALSE)
+write.csv2(per_thing, "flattr-revenue-things.csv")
 
 # summarize & order by month and thing to provide click-value development over time
 per_month_and_thing <- ddply(raw,
                              c("period", "title", "EUR_per_click"),
-                             summarize, all_clicks = sum(clicks),
+                             summarize,
+                             all_clicks = sum(clicks),
                              all_revenue = sum(revenue))
 per_month_and_thing <- per_month_and_thing[order(per_month_and_thing$title),]
-rownames(per_month_and_thing) <- NULL
-write.csv2(per_month_and_thing,
-           "flattr-revenue-clicks.csv",
-           row.names = FALSE)
+write.csv2(per_month_and_thing, "flattr-revenue-clicks.csv")
 
 # summarize & export revenue per month
 per_month <- ddply(raw,
@@ -104,58 +87,63 @@ per_month <- ddply(raw,
                    all_clicks = sum(clicks),
                    all_revenue = sum(revenue))
 per_month <- per_month[order(per_month$period),]
-write.csv2(per_month,
-           "flattr-revenue-months.csv",
-           row.names = FALSE)
+write.csv2(per_month, "flattr-revenue-months.csv")
 
 # revenue per click and month colored by thing, with trends for everything & best thing
 best_thing <- subset(per_month_and_thing, title == per_thing[1,1])  #  reduces data frame to best thing, for later trendline
-rownames(best_thing) <- NULL
 best_thing$EUR_per_click <- best_thing$all_revenue / best_thing$all_clicks
 
-flattr_plot <- ggplot(data = raw, mapping = aes(x = period, y = EUR_per_click,
-                                                size = raw$revenue,  #  points sized according to revenue of that thing in that month => bubble plot
-                                                colour = factor(title)))  +
+flattr_plot <- ggplot(data = raw,
+                      mapping = aes(x = period,
+                                    y = EUR_per_click,
+                                    size = raw$revenue,  #  points sized according to revenue of that thing in that month => bubble plot
+                                    colour = factor(title)))  +
   geom_jitter()  +  # same as geom_point(position = "jitter"); spreads data points randomly around true x value bit; day-exact resolution not (yet) possible
-  labs(list(title = "Development of Flattr Revenue per Click", x = NULL,
-            y = expression("EUR per Flattr (extremes omitted)")))  +  # learned from http://docs.ggplot2.org/current/labs.html
-  stat_smooth(mapping = aes(best_thing$period, best_thing$EUR_per_click, size = best_thing$all_revenue),
-              data = best_thing, method = "auto", show_guide = FALSE, size = N_months / 20,
+  labs(title = "Development of Flattr Revenue per Click",
+       x = NULL, y = expression("EUR per Flattr (extremes omitted)"))  +  # learned from http://docs.ggplot2.org/current/labs.html
+  stat_smooth(mapping = aes(best_thing$period,
+                            best_thing$EUR_per_click,
+                            size = best_thing$all_revenue),
+              data = best_thing,
+              method = "auto",
+              show_guide = FALSE,
+              size = N_months / 20,
               se = FALSE,  #  confidence interval indicator
               linetype = "dashed")  +   # learned from http://sape.inf.usi.ch/quick-reference/ggplot2/linetype
   stat_smooth(aes(group = 1),  # plots trendline over all values; otherwise: one for each thing; learned from http://stackoverflow.com/a/12810890
-              method = "auto", se = FALSE, color = "darkgrey", show_guide = FALSE, size = N_months / 20)  +
-  scale_y_continuous(limits = c(0, mean(raw$EUR_per_click) * 5),  # omit y-values larger than 5x arithmetic mean learned from http://stackoverflow.com/a/26558070
+              method = "auto",
+              se = FALSE,
+              color = "darkgrey",
+              show_guide = FALSE,
+              size = N_months / 20)  +
+  scale_y_continuous(limits = c(0, mean(raw$EUR_per_click) * 5),  # omit extreme y-values; learned from http://stackoverflow.com/a/26558070
                      expand = c(0, 0))  +
-  theme_classic()  +
-  theme(legend.position = "none")
+  theme(legend.position = "none")  +
+  theme_classic(base_size = sqrt(N_things + N_months))
 flattr_plot
-ggsave("flattr-revenue-clicks.png", flattr_plot, limitsize = FALSE)
+ggsave("flattr-revenue-clicks.png", height = N_things/3, width = N_months/1.5)
 
 # revenue per month and thing
-monthly_advanced_plot <- ggplot(per_month_and_thing, aes(x = period, y = all_revenue, fill = factor(title)))  +
+monthly_advanced_plot <- ggplot(per_month_and_thing, aes(period, all_revenue, fill = factor(title)))  +
   geom_bar(stat = "identity")  +
-  labs(list(title = "Development of Flattr Revenue by Things", x = NULL, y = "EUR received"))  +
+  labs(title = "Development of Flattr Revenue by Things", x = NULL, y = "EUR received")  +
   scale_y_continuous(limits = c(0, max(per_month$all_revenue) * 1.1), expand = c(0, 0))  +
   scale_x_date(expand = c(0, 0))  +
-  theme_classic()  +
-  theme(legend.position = "none")
+  theme(legend.position = "none")  +
+  theme_classic(base_size = (N_things + N_months) / 5)
 monthly_advanced_plot
-ggsave("flattr-revenue-months.png", monthly_advanced_plot, limitsize = FALSE)
+ggsave("flattr-revenue-months.png", height = N_things/3, width = N_months/1.5)
 
 # total revenue per month with trend
-monthly_simple_plot <- ggplot(data = per_month, aes(x = period, y = all_revenue))  +
+monthly_simple_plot <- ggplot(per_month, aes(x = period, y = all_revenue))  +
   geom_bar(stat = "identity", group = 1, fill = "#ED8C3B")  +
-  labs(list(title = "Development of Flattr Revenue",
-            y = "EUR received",
-            x = NULL))  +
-  stat_smooth(data = per_month, method = "auto", color = "#80B04A", size = N_months / 5)  +  # fit trend plus confidence interval
-  scale_y_continuous(limits = c(0, max(per_month$all_revenue) * 1.1),  # omit negative y-values & limit positive y-axis to 10% overhead over maximum value
-                     expand = c(0, 0))  +
+  labs(title = "Development of Flattr Revenue", x = NULL, y = "EUR received")  +
+  stat_smooth(data = per_month, method = "auto", color = "#80B04A", size = N_months/5)  +  # fit trend plus confidence interval
+  scale_y_continuous(limits = c(0, max(per_month$all_revenue) * 1.1), expand = c(0, 0))  +
   scale_x_date(expand = c(0, 0))  +
-  theme_classic()
+  theme_classic(base_size = (N_things + N_months) / 10)
 monthly_simple_plot
-ggsave("flattr-revenue-months-summarized.png", monthly_simple_plot, limitsize = FALSE)
+ggsave("flattr-revenue-months-summarized.png")
 
 
 # revenue per location of button
@@ -177,39 +165,29 @@ per_month_and_domain <- ddply(raw,
                               all_clicks = sum(clicks),
                               all_revenue = sum(revenue))
 
-monthly_domain_plot <- ggplot(per_month_and_domain, aes(x = period, y = all_revenue, fill = factor(domain)))  +
+monthly_domain_plot <- ggplot(per_month_and_domain, aes(period, all_revenue, fill = factor(domain))) +
   geom_bar(stat = "identity")  +
-  labs(list(title = "Development of Flattr Revenue by Button Locations",
-            y = "EUR received",
-            x = NULL,
-            fill = "Domains"))  +
+  labs(title = "Development of Flattr Revenue by Button Locations", x = NULL, y = "EUR received", fill = "Domains")  +
   guides(fill = guide_legend(reverse = TRUE))  +
   scale_x_date(expand = c(0,0))  +
-  scale_y_continuous(limits = c(0, max(per_month$all_revenue) * 1.1),
-                     expand = c(0, 0))  +
+  scale_y_continuous(limits = c(0, max(per_month$all_revenue) * 1.1), expand = c(0, 0))  +
   scale_fill_brewer(type = "qual")  +
-  theme_classic()
-  monthly_domain_plot
-ggsave("flattr-revenue-months-domain.png", monthly_domain_plot, limitsize = FALSE)
-
-monthly_domain_plot_fractions <- ggplot(per_month_and_domain,
-                                       aes(period, all_revenue, fill = factor(domain)))  +
-  geom_bar(position = "fill",
-           stat = "identity")  +
-  labs(list(title = "Fractions of Flattr Revenue by Button Locations",
-            x = NULL, y = NULL,
-            fill = "Domains"))  +
+  theme_classic(base_size = (N_things + N_months) / 10)
+monthly_domain_plot
+ggsave("flattr-revenue-months-domain.png")
+
+monthly_domain_plot_fractions <- ggplot(per_month_and_domain, aes(period, all_revenue, fill = factor(domain)))  +
+  geom_bar(position = "fill", stat = "identity")  +
+  labs(title = "Fractions of Flattr Revenue by Button Locations",
+       x = NULL, y = NULL, fill = "Domains")  +
   guides(fill = guide_legend(reverse = TRUE))  +
   scale_x_date(expand = c(0,0))  +
   scale_y_continuous(expand = c(0, 0))  +
   scale_fill_brewer(type = "qual")  +
-  theme_classic()
+  theme_classic(base_size = (N_things + N_months) / 10)
 monthly_domain_plot_fractions
-ggsave("flattr-revenue-months-domain-fractions.png", monthly_domain_plot, limitsize = FALSE)
+ggsave("flattr-revenue-months-domain-fractions.png")
 
 # sort & export after plotting in order to preserve alphabatic sorting in of domains in plot
 per_month_and_domain <- per_month_and_domain[order(per_month_and_domain$all_revenue),]
-rownames(per_month_and_domain) <- NULL
-write.table(per_month_and_domain,
-            "flattr-revenue-clicks-domain.csv",
-            row.names = FALSE)
+write.csv2(per_month_and_domain, "flattr-revenue-clicks-domain.csv")
