1. Сделал подгрузку из ранее подготовленного файла.

2. Посчитал "денежки"

72 bee_cdp/process_bee_cdr.R

@@ -37,6 +37,8 @@ fields <-
     )
   )
 
+# ëèáî ïàðñèì èñõîäíûå ôàéëû, ëèáî ïîäãðóæàåì óæå ðàíåå êîíñîëèäèðîâàííûé
+if (FALSE){
 cdr_spec <- tibble(fields) %>% 
   separate(fields, into=c("cname", "ctype"), sep=":") %>% 
   replace_na(list(ctype = "c"))
@@ -89,6 +91,17 @@ df <- df0 %>%
 
 toc()
 
+tic()
+write_csv(df, "CDP_result.log.gz")
+saveRDS(df, file="CDP_result.rds")
+toc()
+} else{
+  tic()
+  df <- readRDS("CDP_result.rds")  
+  toc()
+}
+
+
 tic()
 # ïîñ÷èòàåì êîëè÷åñòâî óíèêàëüíûõ çíà÷åíèé â êîëîíêàõ
 dist_cols <- map_df(df, n_distinct)
@@ -103,9 +116,6 @@ a_df <- df %>%
   summarise(n=n()) %>%
   arrange(desc(n))
 
-tic()
-write_csv(df, "CDP_result.log.gz")
-toc()
 
 # ïîïðîáóåì îòîáðàçèòü ãðàôèêè. ïðåäâàðèòåëüíàÿ ïîäãîòîâêà ================
 df %<>%
@@ -191,6 +201,24 @@ gg = gg + facet_wrap(~message_type, ncol=1)
 
 gg
 
+# Ïîèññëåäóåì äåíåæêè ====================================
+df3 <- df %>%
+  filter(!is.na(sdr_amount) & sdr_amount>0) %>%
+  group_by(CP) %>%
+  summarize(rur=sum(sdr_amount)/100000)
+
+
+gg <- ggplot(df3 %>% top_n(10, rur), aes(x=fct_reorder(CP, rur, .desc=TRUE), y=rur)) +
+  geom_bar(stat="identity") +
+  theme_ipsum_rc(base_size=14, axis_title_size=12) +
+  theme(axis.text.x = element_text(angle=90)) +
+  geom_text(aes(label=rur), vjust=-0.5, colour="red") +
+  xlab("Êîíòåíò-ïðîâàéäåð") +
+  ylab("Äåíåæêè çà ïåðèîä àíàëèçà, ðóá.") +
+  ggtitle("Àíàëèç äåíåæíûõ ïîòîêîâ")  
+
+gg
+
 if (FALSE){
 # ïðîâåðèì êàêîé èç âàðèàíòîâ ïî îòáðîñó ïîëíîñòüþ NA êîëîíîê áûñòðåå
 f1 <- function(df) {