Commit: 9451dab8a6c050e76bbb740c416f16935a801ff8 - myprojects-hg-reborn (hg)

Commit MetaInfo

• Revisión: 9451dab8a6c050e76bbb740c416f16935a801ff8 (tree)
• Tiempo: 2023-01-19 18:34:03
• Autor: Lorenzo Isella <lorenzo.isella@gmai...>
• Commiter: Lorenzo Isella

Log Message

Almost final form of the script to process the Romanian tam data.

Cambiar Resumen

• modified: R-codes/process_tam_RO.R

Diferencia incremental

diff -r 324cf8383efa -r 9451dab8a6c0 R-codes/process_tam_RO.R
--- a/R-codes/process_tam_RO.R	Wed Jan 18 16:31:22 2023 +0100
+++ b/R-codes/process_tam_RO.R	Thu Jan 19 10:34:03 2023 +0100
@@ -33,22 +33,22 @@
     clean_names()  |> 
     select(-c(text_integral_masura, executanti))
 
-df_name <- read_csv("correspondence.csv") |> 
+df_name <- read_csv("correspondence_modified.csv") |> 
     clean_names()  |> 
     pattern_to_na("...")  |> 
     complete_data()
 
-tam_names <- c("id", "case_reference", "aid_award_created_date", "aid_award_granted_date", 
-"aid_award_published_date", "aid_award_reference", "case_title_original", 
-"case_title_english", "main_procedure_type_code", "is_co_finance", 
-"aid_award_objective", "aid_award_objective_other_english", "aid_award_instrument", 
-"aid_award_instrument_other_english", "beneficiary_name", "beneficiary_name_english", 
-"national_identification", "national_identification_type", "beneficiary_type", 
-"beneficiary_country", "beneficiary_region", "beneficiary_sector", 
-"granted_aid_absolute_eur", "nominal_aid_absolute_eur", "granted_range_eur", 
-"aid_award_ga_original", "aid_award_ga_english", "aid_award_nuts_code", 
-"creator_country", "year", "granted_value_extended_eur", "nominal_value_extended_eur", 
-"is_covid_case")
+## tam_names <- c("id", "case_reference", "aid_award_created_date", "aid_award_granted_date", 
+## "aid_award_published_date", "aid_award_reference", "case_title_original", 
+## "case_title_english", "main_procedure_type_code", "is_co_finance", 
+## "aid_award_objective", "aid_award_objective_other_english", "aid_award_instrument", 
+## "aid_award_instrument_other_english", "beneficiary_name", "beneficiary_name_english", 
+## "national_identification", "national_identification_type", "beneficiary_type", 
+## "beneficiary_country", "beneficiary_region", "beneficiary_sector", 
+## "granted_aid_absolute_eur", "nominal_aid_absolute_eur", "granted_range_eur", 
+## "aid_award_ga_original", "aid_award_ga_english", "aid_award_nuts_code", 
+## "creator_country", "year", "granted_value_extended_eur", "nominal_value_extended_eur", 
+## "is_covid_case")
     ## names(tam)
 
 
@@ -74,7 +74,8 @@
            ) |> 
     mutate(case_reference=select_left_pattern(case_reference, "/")) |> 
     mutate(           is_covid_case=if_else(case_reference %in% covid$case_reference,
-                                            "Yes", "No") )
+                                            "Yes", "No") ) |>
+    mutate(national_identification=as.character(national_identification))
 
 
 
@@ -91,7 +92,7 @@
 
 
 
-aid <- read_csv("aid_type.csv") |> 
+aid <- read_csv("aid_type_modified.csv") |> 
     complete_data()  |> 
     mutate(aid_instrument_rom=tolower(aid_instrument_rom))
 
@@ -99,11 +100,13 @@
                          "întreprindere mică",
                          "întreprindere mijlocie",
                          "întreprindere mare",
-                         "altă categorie de întreprindere"),
+                         "altă categorie de întreprindere",
+                         "imm"),
                    new=c("Small and medium-sized enterprises",
                          "Small and medium-sized enterprises",
                          "Small and medium-sized enterprises",
                          "Only large enterprises",
+                         "",
                          ""
                          ))
 
@@ -119,7 +122,7 @@
     
     select(-c(amount_of_aid_awarded_per_sub_category_in_romanian_lei,obs_value))  |> 
     mutate(beneficiary_sector=as.character(beneficiary_sector)) |> 
-    rename("case_title"="case_title_original",
+    rename(## "case_title"="case_title_original",
            "aid_award_instrument_other_english"="aid_award_instrument") |> 
     mutate(aid_award_instrument_other_english=tolower(aid_award_instrument_other_english))  |> 
     mutate(aid_award_instrument_other_english=recode_many(aid_award_instrument_other_english, aid$aid_instrument_rom ,aid$aid_instrument_eng)) |> 
@@ -146,7 +149,9 @@
   "nominal_aid_absolute_eur", "granted_aid_absolute_eur",
  "beneficiary_region" ,      "beneficiary_sector"      ,
  "case_reference"      ,     "beneficiary_country"     ,
- "year") 
+ "national_identification",
+ "year" , "case_title_original"
+ ) 
 
 ## microîntreprindere	Small and medium-sized enterprises
 ## întreprindere mică	Small and medium-sized enterprises
@@ -166,7 +171,7 @@
     rename("aid_award_instrument"="aid_award_instrument_other_english") |> 
     select(any_of(nn)) |>
     mutate(beneficiary_country="Romania") |>
-    mutate(nominal_value_extended_eur=nominal_aid_absolute_eur,
+    mutate(nominal_value_extended_eur=granted_aid_absolute_eur,
            granted_value_extended_eur=granted_aid_absolute_eur,
            is_covid_case=if_else(case_reference %in% covid$case_reference, "Yes", "No")) |>
     mutate(granted_value_extended_eur=if_else(is_covid_case=="No", granted_value_extended_eur, NA_real_)) |>