Merge pull request #16 from aaronshaw/master

[covid19.git] / wikipedia / example_analysis / revisions_example.R

diff --git a/wikipedia/example_analysis/revisions_example.R b/wikipedia/example_analysis/revisions_example.R
new file mode 100644 (file)
index 0000000..86c7770
--- /dev/null
+++ b/wikipedia/example_analysis/revisions_example.R
@@ -0,0 +1,51 @@
+### COVID-19 Digital Observatory
+### 
+### Minimal example analysis file using revisions data.
+###
+### Note: The revisions data files can get quite big and it may not make
+### sense to try to load them into memory in R. This example file
+### loads the first 1,000 rows of a sample file.
+
+library(tidyverse)
+library(lubridate)
+library(scales)
+
+### Import and cleanup first 1000 rows of one datafile from the observatory
+
+DataURL <-
+  url("https://covid19.communitydata.science/datasets/wikipedia/digobs_covid19-wikipedia-enwiki_revisions-20200401.tsv")
+
+revisions <- read.table(DataURL, sep="\t", header=TRUE,
+                        stringsAsFactors=FALSE,
+                        nrows=1000) 
+
+revisions$timestamp <- as_datetime(revisions$timestamp)
+
+### Group edit by editor to see edit counts
+user.edit.counts <- revisions %>%
+    group_by(user, title) %>% # for the example, there's only one
+                                        # article title
+    summarize(appearances = n()) %>%
+    arrange(-appearances)
+
+### Export that as a little table
+write.table(user.edit.counts,
+            file="output/user_edit_counts-first1000.csv", sep=",",
+            row.names=FALSE)
+
+### A simple time series of edits
+plot <- revisions %>%
+    arrange(timestamp) %>%
+    mutate(count = as.numeric(rownames(revisions))) %>%
+    ggplot(aes(x=timestamp, y=count)) +
+    geom_jitter(alpha=0.1) +
+    xlab("Date") +
+    ylab("Total edits (cumulative)") +
+    ggtitle(paste("Revisions to article: ",
+                  head(revisions$title,1), sep="")) +
+    theme_minimal()
+
+ggsave("output/first_1000_edit_timeseries.png", plot)
+
+
+