Merge branch 'master' into dev

worldbank · May 6, 2019 · 6a09d51 · 6a09d51
2 parents b872c76 + 4e3615b
commit 6a09d51
Show file tree

Hide file tree

Showing 24 changed files with 41,706 additions and 471 deletions.
diff --git a/DataWork/DataSets/DG1__DS_DIR_HDR b/DataWork/DataSets/DG1__DS_DIR_HDR
diff --git a/DataWork/DataSets/Final/DG1__DS_DIR_HDR b/DataWork/DataSets/Final/DG1__DS_DIR_HDR
diff --git a/DataWork/DataSets/Final/whr_panel.csv b/DataWork/DataSets/Final/whr_panel.csv
diff --git a/DataWork/DataSets/Raw/DG1__DS_DIR_HDR b/DataWork/DataSets/Raw/DG1__DS_DIR_HDR
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/DG1__DS_DIR_HDR b/DataWork/DataSets/Raw/UN stats yearbook/DG1__DS_DIR_HDR
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T02_Pop.csv b/DataWork/DataSets/Raw/UN stats yearbook/T02_Pop.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T05_parliament.csv b/DataWork/DataSets/Raw/UN stats yearbook/T05_parliament.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T07_Education.csv b/DataWork/DataSets/Raw/UN stats yearbook/T07_Education.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T08_Teachers.csv b/DataWork/DataSets/Raw/UN stats yearbook/T08_Teachers.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T10_Health_Personnel.csv b/DataWork/DataSets/Raw/UN stats yearbook/T10_Health_Personnel.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T11_HealthExpend.csv b/DataWork/DataSets/Raw/UN stats yearbook/T11_HealthExpend.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T12_Crime.csv b/DataWork/DataSets/Raw/UN stats yearbook/T12_Crime.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T13_Gdp.csv b/DataWork/DataSets/Raw/UN stats yearbook/T13_Gdp.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T17_Unemployme.csv b/DataWork/DataSets/Raw/UN stats yearbook/T17_Unemployme.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/T29_Internet.csv b/DataWork/DataSets/Raw/UN stats yearbook/T29_Internet.csv
diff --git a/DataWork/DataSets/Raw/UN stats yearbook/draft.R b/DataWork/DataSets/Raw/UN stats yearbook/draft.R
@@ -0,0 +1,115 @@
+#------------------------------------------------------------------------------#
+
+#		UN stats year book data processing
+
+#------------------------------------------------------------------------------#
+
+
+
+if (Sys.getenv("USERNAME") == "Leonardo"){
+  projectFolder  <- "C:/Users/Leonardo/Documents/GitHub/dime-r-training"
+  dropbox <- "C:/Users/Leonardo/Dropbox/Work/WB/R Training/R training"
+
+}
+
+if (Sys.getenv("USERNAME") == "WB519128"){
+  projectFolder <- file.path("C:/Users/WB519128/Documents/GitHub/dime-r-training")
+  dropbox <- "C:/Users/WB519128/Dropbox/Work/WB/R Training/R training"
+
+}
+
+UNstatsDat <- file.path(dropbox,"Data/World/Raw/UN stats yearbook") 
+
+
+#------------------------------------------------------------------------------#
+#### Load data ####
+
+pop <- read.csv(file.path(UNstatsDat, "T02_Pop.csv"), header = T, stringsAsFactors = F)
+int <- read.csv(file.path(UNstatsDat,"T29_Internet.csv"), header = T, stringsAsFactors = F)
+crime <- read.csv(file.path(UNstatsDat,"T12_Crime.csv"), header = T, stringsAsFactors = F)
+gdp <- read.csv(file.path(UNstatsDat,"T13_Gdp.csv"), header = T, stringsAsFactors = F)
+educ <- read.csv(file.path(UNstatsDat,"T07_Education.csv"), header = T, stringsAsFactors = F)
+teach <- read.csv(file.path(UNstatsDat,"T08_Teachers.csv"), header = T, stringsAsFactors = F)
+med <- read.csv(file.path(UNstatsDat,"T10_Health_Personnel.csv"), header = T, stringsAsFactors = F)
+
+#------------------------------------------------------------------------------#
+#### Basic cleaning  ####
+
+treat <- 
+  function(x){
+    names(x) <- c("code",
+                  "country",
+                  "year",
+                  "var",
+                  "value",
+                  "coomments",
+                  "source")
+
+    nonCountries <- c("Total, all countries or areas",
+                      "Sub-Saharan Africa",
+                      "Latin America & the Caribbean",
+                      "South-central Asia",
+                      "Australia and New Zealand",
+                      "Oceania",
+                      "Africa",
+                      "Northern Africa",
+                      "Eastern Africa",
+                      "Middle Africa",
+                      "Southern Africa",
+                      "Western Africa",
+                      "Americas",
+                      "Northern America",
+                      "Caribbean",
+                      "Central America",
+                      "South America",
+                      "Asia",
+                      "Central Asia",
+                      "Eastern Asia",
+                      "South-eastern Asia",
+                      "Southern Asia",
+                      "Western Asia",
+                      "Europe",
+                      "Eastern Europe",
+                      "Northern Europe",
+                      "Southern Europe",
+                      "Western Europe",
+                      "Melanesia")
+
+    x <- x[!(x$country %in% nonCountries),]
+    x <- x[x$year > 2014,]
+    return(x)
+  }
+
+
+
+pop <- treat(pop)
+int <- treat(int)
+crime <- treat(crime)
+gdp <- treat(gdp)
+educ <- treat(educ)
+teach <-treat(teach)
+med <- treat(med)
+
+#------------------------------------------------------------------------------#
+#### Population ####
+
+pop$coomments <- NULL
+pop$source <- NULL
+
+keepVarspop <- c("Population mid-year estimates (millions)",
+                 "Sex ratio (males per 100 females)")
+
+pop <- pop[pop$var == keepVarspop,]
+
+
+pop <- spread(pop,
+              key = var,
+              value = value)
+
+
+#names(pop) <- 
+
+#------------------------------------------------------------------------------#
+#### Population ####
+
+
diff --git a/DataWork/DataSets/Raw/Un WHR/DG1__DS_DIR_HDR b/DataWork/DataSets/Raw/Un WHR/DG1__DS_DIR_HDR
diff --git a/DataWork/DataSets/Raw/Un WHR/WHR2015.csv b/DataWork/DataSets/Raw/Un WHR/WHR2015.csv
diff --git a/DataWork/DataSets/Raw/Un WHR/WHR2016.csv b/DataWork/DataSets/Raw/Un WHR/WHR2016.csv
diff --git a/DataWork/DataSets/Raw/Un WHR/WHR2017.csv b/DataWork/DataSets/Raw/Un WHR/WHR2017.csv
diff --git a/DataWork/DataSets/Raw/Un WHR/country_codes.csv b/DataWork/DataSets/Raw/Un WHR/country_codes.csv
@@ -0,0 +1,165 @@
+country,code
+Norway,578
+Denmark,208
+Iceland,352
+Switzerland,756
+Finland,246
+Netherlands,528
+Canada,124
+New Zealand,554
+Sweden,752
+Australia,36
+Israel,376
+Costa Rica,188
+Austria,40
+United States,840
+Ireland,372
+Germany,276
+Belgium,56
+Luxembourg,442
+United Kingdom,826
+Chile,152
+United Arab Emirates,784
+Brazil,76
+Czech Republic,203
+Argentina,32
+Mexico,484
+Singapore,702
+Malta,470
+Uruguay,858
+Guatemala,320
+Panama,591
+France,250
+Thailand,764
+Taiwan,158
+Spain,724
+Qatar,634
+Colombia,170
+Saudi Arabia,682
+Trinidad and Tobago,780
+Kuwait,414
+Slovakia,703
+Bahrain,48
+Malaysia,458
+Nicaragua,558
+Ecuador,218
+El Salvador,222
+Poland,616
+Uzbekistan,860
+Italy,380
+Russia,643
+Belize,84
+Japan,392
+Lithuania,440
+Algeria,12
+Latvia,428
+South Korea,410
+Moldova,498
+Romania,642
+Bolivia,68
+Turkmenistan,795
+Kazakhstan,398
+North Cyprus,196
+Slovenia,705
+Peru,604
+Mauritius,480
+Cyprus,196
+Estonia,233
+Belarus,112
+Libya,434
+Turkey,792
+Paraguay,600
+Hong Kong,344
+Philippines,608
+Serbia,688
+Jordan,400
+Hungary,348
+Jamaica,388
+Croatia,191
+Kosovo,
+China,156
+Pakistan,586
+Indonesia,360
+Venezuela,862
+Montenegro,499
+Morocco,504
+Azerbaijan,31
+Dominican Republic,214
+Greece,300
+Lebanon,422
+Portugal,620
+Bosnia and Herzegovina,70
+Honduras,340
+Macedonia,807
+Somalia,706
+Vietnam,704
+Nigeria,566
+Tajikistan,762
+Bhutan,64
+Kyrgyzstan,417
+Nepal,524
+Mongolia,496
+South Africa,710
+Tunisia,788
+Palestinian Territories,275
+Egypt,818
+Bulgaria,100
+Sierra Leone,694
+Cameroon,120
+Iran,364
+Albania,8
+Bangladesh,50
+Namibia,516
+Kenya,404
+Mozambique,508
+Myanmar,104
+Senegal,686
+Zambia,894
+Iraq,368
+Gabon,266
+Ethiopia,231
+Sri Lanka,144
+Armenia,51
+India,356
+Mauritania,478
+Congo (Brazzaville),178
+Georgia,268
+Congo (Kinshasa),180
+Mali,466
+Ivory Coast,384
+Cambodia,116
+Sudan,729
+Ghana,288
+Ukraine,804
+Uganda,800
+Burkina Faso,854
+Niger,562
+Malawi,454
+Chad,148
+Zimbabwe,716
+Lesotho,426
+Angola,24
+Afghanistan,4
+Botswana,72
+Benin,204
+Madagascar,450
+Haiti,332
+Yemen,887
+South Sudan,728
+Liberia,430
+Guinea,324
+Togo,768
+Rwanda,646
+Syria,760
+Tanzania,834
+Burundi,108
+Central African Republic,140
+Puerto Rico,630
+Suriname,740
+Somaliland Region,
+Laos,418
+Comoros,174
+Oman,512
+Somaliland region,
+Swaziland,748
+Djibouti,262