Einige kleinere Aenderungen

basics.R

@@ -11,13 +11,18 @@ ggplot(sites, aes(score)) +
   labs(title="Verteilung der Punktzahlen", x="Punktzahl", y="Anzahl Sites")
 ggsave("plots/punkte_verteilung.png")
 
+
 ## Verteilung der Punktzahlen (nur KV)
+summary(sites_kv$score)
+
 ggplot(sites_kv, aes(score)) +
   geom_histogram(bins = 15) +
   labs(title="Verteilung der Punktzahlen (nur KV)", x="Punktzahl", y="Anzahl Sites")
 ggsave("plots/punkte_verteilung_kv.png")
 
 ## Verteilung der Punktzahlen (nur OV)
+summary(sites_ov$score)
+
 ggplot(sites_ov, aes(score)) +
   geom_histogram(bins = 15) +
   labs(title="Verteilung der Punktzahlen (nur OV)", x="Punktzahl", y="Anzahl Sites")
@@ -66,6 +71,43 @@ ggplot(sites_top_cms, aes(generator)) +
        y="Anzahl Sites")
 ggsave("plots/feeds_nach_cms_b.png")
 
-# Feeds - Nur GCMS
-ggplot(sites_generator_gcms, aes(rating.FEEDS.value)) +
+
+## Alle Kriterien
+
+sites_rating_criteria <- select(sites,
+                                rating.DNS_RESOLVABLE_IPV4.value,
+                                rating.SITE_REACHABLE.value,
+                                rating.CANONICAL_URL.value,
+                                rating.WWW_OPTIONAL.value,
+                                rating.HTTPS.value,
+                                rating.RESPONSIVE.value,
+                                rating.CONTACT_LINK.value,
+                                rating.SOCIAL_MEDIA_LINKS.value,
+                                rating.NO_NETWORK_ERRORS.value,
+                                rating.NO_SCRIPT_ERRORS.value,
+                                rating.NO_THIRD_PARTY_COOKIES.value,
+                                rating.FAVICON.value,
+                                rating.USE_SPECIFIC_FONTS.value,
+                                rating.FEEDS.value,
+)
+sites_rating_criteria_long <- tidyr::gather(sites_rating_criteria, key = type_col, value = categories)
+
+sapply(sites_rating_criteria_long,table)
+
+# requires library(qdapTools)
+mtabulate(sites_rating_criteria_long)
+
+(sum(sites_rating_criteria$rating.FEEDS.value) / count(sites)) * 100
+(sum(sites_rating_criteria$rating.SOCIAL_MEDIA_LINKS.value) / count(sites)) * 100
+(sum(sites_rating_criteria$rating.RESPONSIVE.value) / count(sites)) * 100
+(sum(sites_rating_criteria$rating.NO_THIRD_PARTY_COOKIES.value) / count(sites)) * 100
+
+
+ggplot(sites_rating_criteria_long, aes(x = categories, fill = categories)) +
+  geom_bar() + 
+  coord_flip() +
+  facet_wrap(~ type_col, scales = "free_x")
+
+ggplot(sites_rating_criteria_long, aes(x = categories, fill = categories)) +
   geom_bar()
+

common.R

@@ -2,6 +2,7 @@ library(jsonlite)
 library(dplyr)
 library(ggplot2)
 library(scales)
+library(tidyr)
 
 # Frische Daten gibt es per
 # curl https://green-spider.netzbegruenung.de/api/v1/spider-results/table/ > data/table.json

response_duration.R

@@ -20,7 +20,7 @@ ggplot(sites, aes(rating.HTTP_RESPONSE_DURATION.value)) +
   labs(title="Verteilung der Antwortzeiten", x="Antwortzeit in Millisekunden (Median in blau)", y="Häufigkeit")
 ggsave("plots/antwortzeiten_verteilung_kurve.png")
 
-## Verteilung der Punkte (logarithmische X-Achse)
+## Verteilung der Antwortzeiten (logarithmische X-Achse)
 ggplot(sites, aes(rating.HTTP_RESPONSE_DURATION.value)) +
   geom_histogram() +
   scale_x_log10() +