Revision | ab996ce51119a17fea8fcf85809c20f82497b855 (tree) |
---|---|
Time | 2020-05-12 06:26:08 |
Author | Lorenzo Isella <lorenzo.isella@gmai...> |
Commiter | Lorenzo Isella |
I added a file to process the hs0 data.
@@ -0,0 +1,47 @@ | ||
1 | +rm(list=ls()) | |
2 | + | |
3 | + | |
4 | + | |
5 | +library(tidyverse) | |
6 | +library(janitor) | |
7 | + | |
8 | +source("/home/lorenzo/myprojects-hg/R-codes/stat_lib.R") | |
9 | + | |
10 | + | |
11 | +## remove_last_weeks <- 3 | |
12 | + | |
13 | +## aver_on_latest_weeks <- 4 | |
14 | + | |
15 | +iso_eu27 <- c("AT", "BE", "BG", "CY", "CZ", | |
16 | + "DK", "EE", "FI", "FR", "DE", "GR", "HU", "IE", | |
17 | + "IT", "LV", "LT", "LU", "MT", "NL", "PL", | |
18 | + "PT", "RO", "SK", "SI", "ES", "SE", "HR") | |
19 | + | |
20 | + | |
21 | + | |
22 | + | |
23 | +df_ini <- read_delim("cdc_hs0.txt", delim=";", na="")%>% | |
24 | + clean_names() %>% | |
25 | + mutate(year=as.integer(year), week=as.integer(week), | |
26 | + iv=iv/1e6) %>% | |
27 | + select(-iq, -productid) | |
28 | + | |
29 | + | |
30 | + | |
31 | +df <- df_ini %>% | |
32 | + filter(partnerid!="EU") %>% | |
33 | + group_by(year, week, partnerid) %>% | |
34 | + group_modify(~ bind_rows(.x,tibble(reporterid="EU27", | |
35 | + iv=sum(.x$iv[.x$reporterid %in% iso_eu27 ], na.rm=T)))) %>% | |
36 | + ungroup %>% | |
37 | + group_by(year, week, reporterid) %>% | |
38 | + group_modify(~ bind_rows(.x,tibble(partnerid="Extra-EU28", | |
39 | + iv=sum(.x$iv, na.rm=T)))) %>% | |
40 | + ungroup | |
41 | + | |
42 | + | |
43 | +saveRDS(df, "./app_hs0/data_hs0.RDS") | |
44 | +saveRDS(df, "data_hs0.RDS") | |
45 | + | |
46 | + | |
47 | +print("So far so good") |