Meilleure désambiguification des n°RSS par établissement

donnees.R

@@ -59,9 +59,9 @@ unlink(str_c("OVALIDE T2A.MCO.DGF", annee - 1, mois, "html", sep = "."))
 # RUM + structure + calcul durée de séjour
 print("LIVENNE : RUM et structure")
 src_mysql("pmsi_dim", "localhost", user = user, password = password) %>%
-  tbl(sql("SELECT idhosp, idrss, idrum, date_entree, date_sortie, cmd, um FROM fix116
+  tbl(sql("SELECT finess, idhosp, idrss, idrum, date_entree, date_sortie, cmd, um FROM fix116
           UNION
-          SELECT idhosp, idrss, idrum, date_entree, date_sortie, cmd, um FROM fix117")) %>%
+          SELECT finess, idhosp, idrss, idrum, date_entree, date_sortie, cmd, um FROM fix117")) %>%
   collect %>%
   mutate(date_entree = date_entree %>% as.Date,
          date_sortie = date_sortie %>% as.Date) %>%
@@ -77,7 +77,9 @@ src_mysql("pmsi_dim", "localhost", user = user, password = password) %>%
               filter(dummy == last(dummy)) %>%
               select(-dummy, -uf_date_ferm) %>%
             ungroup) %>%
-  mutate(idrss = ifelse(idhosp %>% as.numeric < 500000000, str_c("m", idrss), idrss)) %>% # Dédoublonnage rss maternité 2013
+  # mutate(idrss = ifelse(idhosp %>% as.numeric < 500000000, str_c("m", idrss), idrss)) %>% # Dédoublonnage rss maternité 2013
+  mutate(idrss = ifelse(finess == "540000031", str_c("m", idrss), idrss)) %>% # maternité
+  mutate(idrss = ifelse(finess == "540020112", str_c("c", idrss), idrss)) %>% # CCEG
   mutate(idhosp = idhosp %>% str_replace_all(" ", "")) %>%
   mutate(duree_rum = date_sortie - date_entree) %>% # durée rum
   left_join(group_by(., idrss) %>%