diff --git a/.gitignore b/.gitignore index f26b794..7a49d03 100644 --- a/.gitignore +++ b/.gitignore @@ -49,3 +49,4 @@ po/*~ # RStudio Connect folder rsconnect/ +/.idea/ diff --git a/data_extraction_statcan.rmd b/data_extraction_statcan.rmd new file mode 100644 index 0000000..3a3f149 --- /dev/null +++ b/data_extraction_statcan.rmd @@ -0,0 +1,72 @@ +--- +title: "Revenu universel de base avec le retrait de l'exemption fiscale de base" +output: html_document +--- + +## Obtenir les informations pertinentes sur les revenus canadiens depuis Statistiques Canada + +```{r} +library(statcanR) +library(dplyr) +library(stringr) +library(readr) +``` + +```{r} +# Population by income group +population_by_income_group <- statcanR::statcan_download_data(tableNumber="98-10-0064-01", lang="eng") +``` +Cleaning total income column names + +```{r} +names(population_by_income_group) <- c("REF_DATE","GEO","DGUID","AGE","GENDER","TOTAL_INCOME_GROUP", "COORDINATE","Y2020","SYMBOL0","Y2015","SYMBOL","INDICATOR","COORDINATE0") +``` + + +Cleaning total income population data + +```{r} +population_by_income_group %>% + filter(GEO=="Canada", + AGE=="Total - Age", + GENDER=="Total - Gender") %>% + select(REF_DATE,TOTAL_INCOME_GROUP,Y2020,Y2015)-> + population_by_income_group_filtered +``` + + +```{r} +# Population by after-tax income group +population_by_after_tax_income_group <- statcanR::statcan_download_data(tableNumber="98-10-0065-01", lang="eng") +``` +Cleaning after tax column names + +```{r} +names(population_by_after_tax_income_group) <- c("REF_DATE","GEO","DGUID","AGE","GENDER","AFTERTAX_INCOME_GROUP", "COORDINATE","Y2020","SYMBOL0","Y2015","SYMBOL","INDICATOR","COORDINATE0") +``` + +Cleaning after tax population data + +```{r} +population_by_after_tax_income_group %>% + filter(GEO=="Canada", + AGE=="Total - Age", + GENDER=="Total - Gender") %>% + select(REF_DATE,AFTERTAX_INCOME_GROUP,Y2020,Y2015)-> + population_by_after_tax_income_group_filtered +``` + +Save data to csv file + +```{r} +population_by_income_group_filtered %>% + write_csv("population_by_income_group_filtered.csv") +population_by_after_tax_income_group_filtered %>% + write_csv("population_by_after_tax_income_group_filtered.csv") +``` + + +```{r} +# Tax brackets +federal_income_brackets <- read.csv("federal-income-brackets.csv") +``` diff --git a/federal-income-brackets.csv b/federal-income-brackets.csv new file mode 100644 index 0000000..5a98d4a --- /dev/null +++ b/federal-income-brackets.csv @@ -0,0 +1,6 @@ +lower,upper,percent +0,55867,15 +55867,111733,20.5 +111733,173205,26 +173205,246752,29 +246752,10000000,33 \ No newline at end of file diff --git a/population_by_after_tax_income_group_filtered.csv b/population_by_after_tax_income_group_filtered.csv new file mode 100644 index 0000000..936e183 --- /dev/null +++ b/population_by_after_tax_income_group_filtered.csv @@ -0,0 +1,23 @@ +REF_DATE,AFTERTAX_INCOME_GROUP,Y2020,Y2015 +2021-01-01,Total - After-tax income,30336135,28642980 +2021-01-01,Without after-tax income,1079955,1143795 +2021-01-01,With after-tax income,29256180,27499185 +2021-01-01,Percentage with after-tax income,96.4,96 +2021-01-01,"Under $5,000 (including loss)",1461745,2126645 +2021-01-01,"$5,000 to $9,999 ",1099950,1578610 +2021-01-01,"$10,000 to $14,999",1522855,2099155 +2021-01-01,"$15,000 to $19,999",2085400,2269120 +2021-01-01,"$20,000 to $24,999",2805875,2373320 +2021-01-01,"$25,000 to $29,999",2488625,2022355 +2021-01-01,"$30,000 to $34,999",2293055,1878515 +2021-01-01,"$35,000 to $39,999",2149915,1833865 +2021-01-01,"$40,000 to $44,999",2014240,1711765 +2021-01-01,"$45,000 to $49,999",1784470,1504005 +2021-01-01,"$50,000 to $54,999",1530410,1286085 +2021-01-01,"$55,000 to $59,999",1302715,1084450 +2021-01-01,"$60,000 to $69,999",2020895,1703000 +2021-01-01,"$70,000 to $79,999",1428515,1210700 +2021-01-01,"$80,000 to $89,999",1027020,872865 +2021-01-01,"$90,000 to $99,999",643700,547620 +2021-01-01,"$100,000 and over",1596795,1397105 +2021-01-01,Median after-tax income ($),36800,33600 diff --git a/population_by_income_group_filtered.csv b/population_by_income_group_filtered.csv new file mode 100644 index 0000000..7c96e68 --- /dev/null +++ b/population_by_income_group_filtered.csv @@ -0,0 +1,25 @@ +REF_DATE,TOTAL_INCOME_GROUP,Y2020,Y2015 +2021-01-01,Total - Total income,30336135,28642980 +2021-01-01,Without total income,1094320,1154455 +2021-01-01,With total income,29241810,27488530 +2021-01-01,Percentage with total income,96.4,96 +2021-01-01,"Under $5,000 (including loss)",1372840,2046650 +2021-01-01,"$5,000 to $9,999",1061890,1540875 +2021-01-01,"$10,000 to $14,999",1482780,2059750 +2021-01-01,"$15,000 to $19,999",1965405,2148060 +2021-01-01,"$20,000 to $24,999",2509900,2109215 +2021-01-01,"$25,000 to $29,999",2125755,1698010 +2021-01-01,"$30,000 to $34,999",1922550,1542910 +2021-01-01,"$35,000 to $39,999",1801130,1480060 +2021-01-01,"$40,000 to $44,999",1712405,1438020 +2021-01-01,"$45,000 to $49,999",1578315,1344745 +2021-01-01,"$50,000 to $54,999",1423630,1200385 +2021-01-01,"$55,000 to $59,999",1253835,1056840 +2021-01-01,"$60,000 to $69,999",2073335,1751840 +2021-01-01,"$70,000 to $79,999",1592465,1351425 +2021-01-01,"$80,000 to $89,999",1239395,1078305 +2021-01-01,"$90,000 to $99,999",965945,821840 +2021-01-01,"$100,000 and over",3160245,2819590 +2021-01-01,"$100,000 to $149,999",2102630,1874030 +2021-01-01,"$150,000 and over",1057615,945555 +2021-01-01,Median total income ($),41200,37200 diff --git a/revenu-universel-exemption-impots.Rproj b/revenu-universel-exemption-impots.Rproj new file mode 100644 index 0000000..8e3c2eb --- /dev/null +++ b/revenu-universel-exemption-impots.Rproj @@ -0,0 +1,13 @@ +Version: 1.0 + +RestoreWorkspace: Default +SaveWorkspace: Default +AlwaysSaveHistory: Default + +EnableCodeIndexing: Yes +UseSpacesForTab: Yes +NumSpacesForTab: 2 +Encoding: UTF-8 + +RnwWeave: Sweave +LaTeX: pdfLaTeX