Clean-up covid data in plan.

src/Covid19MortalityEurope.Rmd

@@ -114,13 +114,29 @@ Lock <- do.call(
        })
 )
 
+## Sanity checks in covid 
+## there is exactly one record per country and date
+# covid %>% group_by(date, name) %>% count() %>% pull(n) %>% identical(rep(1L, length(.)))  # TRUE
+## The population is consistent for each country
+# covid %>% 
+#   count(name, pop) %>% 
+#   count(name, wt = 1L) %>% 
+#   pull(n) %>% 
+#   identical(rep(1L, length(.)))
+
 deaths_by_ctry_dt <- covid %>%
-    filter(deaths >= 0) %>% 
     group_by(date, name) %>%
-    summarise(deaths = sum(deaths)) %>%
-    ungroup() %>%
+    summarise(deaths = sum(deaths), .groups = "drop") %>%
     mutate(date = as.Date(date))
 
+# identical(
+#   deaths_by_ctry_dt,
+#   covid %>% 
+#     select(date, name, deaths) %>% 
+#     mutate(date = as.Date(date)) %>% 
+#     arrange(date, name)
+# )
+
 data_update_date <- format(Sys.Date() - 1,
                            "%d %b %Y")
 

src/plan.R

@@ -63,7 +63,10 @@ plan <- drake_plan(## raw_data = read_excel(
     sdt = format(Sys.Date(),
                  "%d %b %Y"),
 
-    covid = ecdc_covid19(Sys.Date() - 1),
+    covid = ecdc_covid19(Sys.Date() - 1) %>%
+        # rm some negative records that are obvious errors
+        # should we use NA instead?
+        filter(deaths >= 0),
 
     ## ## String notice for data update
     ## data_update_date = format(Sys.Date() - 1,