Make separate examples proper R-style demos

Change-Id: I0ac284cfc1d0c508030c91189f260299680c1485

Author: kupietz

.Rbuildignore

@@ -1,4 +1,3 @@
 ^.*\.Rproj$
 ^\.Rproj\.user$
 ^Readme.md
-^examples/

Readme.md

@@ -27,9 +27,9 @@ library(RKorAPClient)
 kqo <- corpusQuery(new("KorAPConnection", verbose=TRUE), "Hello world")
 fetchAll(kqo)
 ```
-## Examples
+## Demos
 
-More elaborate R scripts demonstrating the use of the package can be found in the [examples](examples) folder.
+More elaborate R scripts demonstrating the use of the package can be found in the [demo](demo) folder.
 
 ## Development and License
 

demo/00Index

@@ -0,0 +1,5 @@
+frequenciesOverTime       Plot frequency of query expressions over time
+frequenciesOverDomains    Box plot frequency of query expressions per topic domain
+conditionsOverTime        Plot frequency of query expressions over time under different conditions
+alternativesOverTime      Plot proportion of alternative spellings/variants over time
+regional                  Map plot regional frequencies of query expression

examples/simple/alternativesOverTime.R renamed to demo/alternativesOverTime.R

@@ -26,7 +26,7 @@ conditionsOverTime <- function(query, conditions, years, kco = new("KorAPConnect
     geom_ribbon(aes(ymin=ci[, 1], ymax=ci[, 2], fill=condition, color=condition), alpha=.3, linetype=0) +
     xlab("TIME") +
     labs(color="Virtual Corpus", fill="Virtual Corpus") +
-    ylab(sprintf("Observed frequency of “%s”", query)) +
+    ylab(sprintf("Observed frequency of \u201c%s\u201d", query)) +
     theme(axis.text.x = element_text(angle = 45, hjust = 1))  + scale_x_continuous(breaks=unique(df$year))
   print(g)
   # print(ggplotly(g, tooltip = c("x", "y")))

examples/simple/conditionsOverTime.R renamed to demo/conditionsOverTime.R

@@ -21,7 +21,7 @@ freqPerDomain <- function(query, con = new("KorAPConnection", verbose = TRUE)) {
   g <- ggplot(data = df, mapping = aes(x = Domain, y = freq)) +
     geom_col() +
     geom_errorbar(aes(ymin=ci[, 1], ymax=ci[, 2]), width=.5, alpha=.5) +
-    ylab(sprintf("Observed frequency of “%s”", query)) +
+    ylab(sprintf("Observed frequency of \u201c%s\u201d", query)) +
     theme(axis.text.x = element_text(angle = 45, hjust = 1))
   print(g)
   df

examples/geo/data/regions.rds renamed to demo/data/regions.rds

@@ -0,0 +1,35 @@
+#!/usr/bin/env Rscript
+#
+# Plot frequency of query expressions over time
+#
+library(RKorAPClient)
+library(ggplot2)
+
+freqPerYear <- function(query, con = new("KorAPConnection", verbose = TRUE)) {
+  vc <- "pubDate since 2000 & pubDate until 2018 & textType = /Zeit.*/"
+  q <- corpusQuery(con, query = query, vc=vc)
+  q <- fetchAll(q)
+  tokensPerYear <- function(year) {
+    return(corpusStats(con, sprintf("%s & pubDate in %s", vc, year))@tokens)
+  }
+  df <- as.data.frame(table(as.numeric(format(q@collectedMatches$pubDate,"%Y")), dnn="year"),
+                      stringsAsFactors = FALSE)
+  df <- merge(data.frame(year=min(df$year):max(df$year)), df, all = TRUE)
+  df[is.na(df$Freq),]$Freq <- 0
+  df$total <- sapply(df$year, tokensPerYear)
+  df$freq <- df$Freq / df$total
+  df$ci <- t(sapply(Map(prop.test, df$Freq, df$total), "[[","conf.int"))
+  g <- ggplot(data = df, aes(x = year, y = freq, group=1)) +
+    geom_ribbon(aes(ymin=ci[, 1], ymax=ci[, 2]), alpha=.3) +
+    geom_point() +
+    geom_line() +
+    xlab("TIME") +
+    ylab(sprintf("Observed frequency of \u201c%s\u201d", query)) +
+    theme(axis.text.x = element_text(angle = 45, hjust = 1))
+  print(g)
+  df
+}
+#df <- freqPerYear("Car-Bikini")
+#df <- freqPerYear("[tt/p=ART & opennlp/p=ART] [tt/l=teilweise] [tt/p=NN]")
+df <- freqPerYear("Buschzulage")
+

examples/simple/frequenciesOverDomains.R renamed to demo/frequenciesOverDomains.R

@@ -6,7 +6,8 @@ library(broom)
 library(plotly)
 library(htmlwidgets)
 
-mapfile <- "examples/geo/data/cache/map-v2.rds"
+devAskNewPage(ask = FALSE)
+mapfile <- "demo/data/cache/map-v2.rds"
 
 fetchAndPrepareMap <- function(map, pick) {
   cat("Downloading GADM map data for ", map, "\n")
@@ -34,7 +35,7 @@ fetchMaps <- function(maps, picks) {
 map <- fetchMaps(c("DEU_1", "AUT_0", "CHE_0", "LUX_0", "BEL_3", "ITA_1", "LIE_0"), c(0, 0, 0, 0, 34, 17, 0))
 
 geoDistrib <- function(query, kco = new("KorAPConnection", verbose=TRUE)) {
-  regions <- readRDS("examples/geo/data/regions.rds")
+  regions <- readRDS("demo/data/regions.rds")
   regions$freq <- NA
   regions$url <- NA
   plot <- NULL
@@ -84,7 +85,7 @@ updatePlot <- function(query, map, regions) {
           axis.title.x=element_blank(),
           axis.title.y=element_blank()) +
     coord_equal(ratio=1.5) +
-    labs(title = sprintf("Regional distribution of “%s”", query))
+    labs(title = sprintf("Regional distribution of \u201c%s\u201d", query))
   print(regionsPlot)
   regionsPlot
 }