Important packages/libraries for this lab:
I. Loading the data set in R.
III. Looking at the data: Describe, Browse, Summarize, Tabulate
- 7.
- 8. Counting number of occurence of a variable
IV. Generating new variables; use mutate to add a new variable to the dataset
V. Creating Tables and Figures to describe economic trends
Visualizing unemployment accross years(2020-2023)

Important packages/libraries for this lab:

tidyverse
foreign: used when reading data that is not in .CSV format
readstata13 : used when reading data in dta format

library(tidyverse)

## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.2 ──
## ✔ ggplot2 3.5.1     ✔ purrr   1.0.2
## ✔ tibble  3.2.1     ✔ dplyr   1.1.4
## ✔ tidyr   1.2.1     ✔ stringr 1.4.1
## ✔ readr   2.1.2     ✔ forcats 1.0.0
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()

if(!require("tufte")) install.packages("tufte")

## Loading required package: tufte

#library(tufte)

if(!require("foreign")) install.packages("foreign")

## Loading required package: foreign

library(foreign) #this will help when importing data that are not in csv.format(stata, sas, spss)
#library(haven)

if(!require("readstata13")) install.packages("readstata13")

## Loading required package: readstata13

library(readstata13)

rm(list = ls()) #clear the environment

I. Loading the data set in R.

We use function “rad.dta13” from the library “readstata13” which read the dta version of dataset from stata version 13 and above

Cps20 <- read.dta13("CPS_2020_monthly.dta")
Cps21<-read.dta13("CPS_2021_23_monthly.dta")

III. Looking at the data: Describe, Browse, Summarize, Tabulate

7.

#quickly look at data structure
glimpse(Cps20) 

#Summary statistic of all variables in the dataset
summary(Cps20) 

prop.table(table(Cps20$month)) # to check the frequency distribution of the value of a variable of interest; shows proportion.
table(Cps20$month)#gives number of occurrence

head(Cps20$month) # see a specific variable from the dataset

is.na(Cps20$unemployed) # check missing data from a variable (return a true or false)
sum(is.na(Cps20$unemployed)) # find count of missing values
which(is.na(Cps20$unemployed)) #find the location of missing variable

8. Counting number of occurence of a variable

# tabulate month
Cps20 %>%
  count(month)%>%# we use count function
  mutate(Percent=round(n/sum(n),4)*100)%>% # frequency rounded at 3
  arrange(desc(Percent))%>% # arranged in descending order
  mutate(Percentile = cumsum(Percent))

##        month     n Percent Percentile
## 1   february 80585    9.33       9.33
## 2    january 79880    9.25      18.58
## 3    october 77294    8.95      27.53
## 4   november 76243    8.83      36.36
## 5  september 75326    8.72      45.08
## 6   december 73545    8.52      53.60
## 7      march 71215    8.25      61.85
## 8      april 68734    7.96      69.81
## 9     august 67169    7.78      77.59
## 10       may 66201    7.67      85.26
## 11      july 64312    7.45      92.71
## 12      june 63003    7.30     100.01

IV. Generating new variables; use mutate to add a new variable to the dataset

10.

# tab to view dummy variable unemployed created
Cps20%>%
  group_by(empstat)%>% #group by empstat
  summarise(mean_unemployed = mean(unemployed),
            n =n()) # compute the mean of unemployed

## # A tibble: 8 × 3
##   empstat                        mean_unemployed      n
##   <fct>                                    <dbl>  <int>
## 1 armed forces                                NA   4296
## 2 at work                                      0 528302
## 3 has job, not at work last week               0  23995
## 4 unemployed, experienced worker               1  41237
## 5 unemployed, new worker                       1   1892
## 6 nilf, unable to work                        NA  45829
## 7 nilf, other                                 NA 143806
## 8 nilf, retired                               NA  74150

11. create a dummy variable female

Cps20<-Cps20 %>%
  mutate(female=ifelse(sex=='female',1,0))

12. Compute the mean of the female variable

# tabulate female
Cps20%>%
  count(female) %>%
  mutate(percent = round(n/sum(n),4)*100)

##   female      n percent
## 1      0 421602   48.82
## 2      1 441905   51.18

# summarise female
Cps20%>%
  summarise(Obs =n(),
            mean_female = round(mean(female),3),
            Std.dev. =round(sd(female),3),
            min = min(female),
            max = max(female))

##      Obs mean_female Std.dev. min max
## 1 863507       0.512      0.5   0   1

# optional: calculating average unemployment rate by sex in each month
Cps20 %>% 
  group_by(month,sex) %>%
  summarise(aver_unemployed =round(mean(unemployed, na.rm =TRUE),3))%>%
  pivot_wider(names_from = sex,
              values_from = aver_unemployed)%>%
  ungroup()

## `summarise()` has grouped output by 'month'. You can override using the
## `.groups` argument.

## # A tibble: 12 × 3
##    month      male female
##    <fct>     <dbl>  <dbl>
##  1 january   0.042  0.035
##  2 february  0.042  0.032
##  3 march     0.046  0.04 
##  4 april     0.124  0.146
##  5 may       0.109  0.129
##  6 june      0.096  0.108
##  7 july      0.087  0.101
##  8 august    0.075  0.08 
##  9 september 0.071  0.071
## 10 october   0.063  0.059
## 11 november  0.062  0.056
## 12 december  0.062  0.057

13. Generate a dummy variable ‘college’ for at least having 4-year college degree.

Create a dummy variable “college”, which is equal to 1 if educ ( a person) has a bachelor’s degree, master’s degree, or a doctorate degree we use case_when function when we are treating more than one condition ( using more than one if-statement)

Cps20<-Cps20 %>%
  mutate(college =case_when(Cps20$educ=="bachelor's degree"~1,
                           Cps20$educ=="master's degree"~1,
                           Cps20$educ=="doctorate degree"~1,
                           TRUE~0))

V. Creating Tables and Figures to describe economic trends

14. Compute unemployment rate accross months

In stata, it’s only one line of code: tab month, sum(unemployed)

# number of employment and unemployment in each month: 
table(Cps20$month, Cps20$unemployed)# tab month unemployed.

##            
##                 0     1
##   january   53695  2152
##   february  54488  2103
##   march     47437  2137
##   april     39895  6217
##   may       39600  5309
##   june      38973  4414
##   july      40456  4169
##   august    42957  3598
##   september 48097  3674
##   october   50043  3284
##   november  49226  3083
##   december  47430  2989

# or you can do it in this way:
unemployment<-Cps20%>%
  filter(!is.na(month)& !is.na(unemployed))%>%
  count(month, unemployed)%>%
  pivot_wider(names_from = unemployed,
              values_from = n) %>%
  rename(unemployment_1="1",
         employment_0="0")
unemployment

## # A tibble: 12 × 3
##    month     employment_0 unemployment_1
##    <fct>            <int>          <int>
##  1 january          53695           2152
##  2 february         54488           2103
##  3 march            47437           2137
##  4 april            39895           6217
##  5 may              39600           5309
##  6 june             38973           4414
##  7 july             40456           4169
##  8 august           42957           3598
##  9 september        48097           3674
## 10 october          50043           3284
## 11 november         49226           3083
## 12 december         47430           2989

#Mean of unemployment rates in each month 
Cps20%>%
  group_by(month)%>%
  #filter(month=="january")%>%
  summarise(Obs = n(),
            mean_unemployed = round(mean(unemployed, na.rm = TRUE), 3),
            Std.dev. = round(sd(unemployed, na.rm=TRUE),3),
            min =min(unemployed, na.rm = TRUE),
            max =max(unemployed, na.rm = TRUE))

## # A tibble: 12 × 6
##    month       Obs mean_unemployed Std.dev.   min   max
##    <fct>     <int>           <dbl>    <dbl> <dbl> <dbl>
##  1 january   79880           0.039    0.192     0     1
##  2 february  80585           0.037    0.189     0     1
##  3 march     71215           0.043    0.203     0     1
##  4 april     68734           0.135    0.342     0     1
##  5 may       66201           0.118    0.323     0     1
##  6 june      63003           0.102    0.302     0     1
##  7 july      64312           0.093    0.291     0     1
##  8 august    67169           0.077    0.267     0     1
##  9 september 75326           0.071    0.257     0     1
## 10 october   77294           0.062    0.24      0     1
## 11 november  76243           0.059    0.236     0     1
## 12 december  73545           0.059    0.236     0     1

 #*To compute the mean at a specific month, use filter function.*

15. Unemployment Across Gender

sum_female<-Cps20%>% 
  group_by(month,female) %>%
  summarise(aver_unemployed = mean(unemployed, na.rm =TRUE))%>%
  pivot_wider(names_from = female,
              values_from = aver_unemployed)%>%
  rename(unemp_male ="0",
         unemp_female="1")

## `summarise()` has grouped output by 'month'. You can override using the
## `.groups` argument.

sum_female

## # A tibble: 12 × 3
## # Groups:   month [12]
##    month     unemp_male unemp_female
##    <fct>          <dbl>        <dbl>
##  1 january       0.0420       0.0348
##  2 february      0.0415       0.0324
##  3 march         0.0462       0.0397
##  4 april         0.124        0.146 
##  5 may           0.109        0.129 
##  6 june          0.0964       0.108 
##  7 july          0.0867       0.101 
##  8 august        0.0746       0.0802
##  9 september     0.0705       0.0715
## 10 october       0.0635       0.0595
## 11 november      0.0617       0.0559
## 12 december      0.0617       0.0566

16. Plots average unemployment accross gender

Plotting using R is optional,You can plot that in excel by using sum_female

ggplot(sum_female)+
  geom_line(aes(month, unemp_male, group=1, color="male" ))+
  geom_line(aes(month, unemp_female, group=1, color="female"))+
  labs(
    title = "Average unemployment rate accross gender",
    subtitle = "in 2020",
    x="Month",y="gender",
    color=""
  )

17. Generate age categories from age variable

Unemployment accross age group

Cps20_1%>%
  group_by(month, agecat)%>%
  summarise(aver_unemp=mean(unemployed, na.rm = TRUE))

## `summarise()` has grouped output by 'month'. You can override using the
## `.groups` argument.

## # A tibble: 72 × 3
## # Groups:   month [12]
##    month    agecat aver_unemp
##    <fct>    <chr>       <dbl>
##  1 january  0          0.129 
##  2 january  1          0.0644
##  3 january  2          0.0346
##  4 january  3          0.0288
##  5 january  4          0.0279
##  6 january  5          0.0282
##  7 february 0          0.111 
##  8 february 1          0.0594
##  9 february 2          0.0340
## 10 february 3          0.0295
## # ℹ 62 more rows

Graph unemployment accross age group

age_graph<-Cps20_1%>%
  group_by(month, agecat)%>%
  summarise(aver_unemp=mean(unemployed, na.rm = TRUE))%>%
  pivot_wider(names_from = agecat,
              values_from = aver_unemp)%>%
  rename(cat0 ="0", cat1="1", cat2="2",cat3="3", cat4="4", cat5="5")%>%
  
  ggplot()+
  geom_line(aes(month, cat0, group=1, color="age<20"))+
  geom_line(aes(month, cat1, group=1, color="between age 20 &30"))+
  geom_line(aes(month, cat2, group=1, color="between age 30 &40"))+
  geom_line(aes(month, cat3, group=1, color="between age 40 &50"))+
  geom_line(aes(month, cat5, group=1, color="between age 50 &60"))+
  geom_line(aes(month, cat5, group=1, color="age>60"))+
  labs(
    title = "Unemployment rates accross age category: 2020",
    x="Months", y="Unemploymnt_rates",
    color ="Age Groups"
  )+
  theme(axis.text.x = element_text(angle = 45, hjust = 0.1, vjust = 0))

## `summarise()` has grouped output by 'month'. You can override using the
## `.groups` argument.

 age_graph

18. Merge two datasets: append(merge across observation)

Cps20_23<-bind_rows(Cps20, Cps21)

Average unemployment accross years (2020-2023):table

Cps20_23%>%
  group_by(year)%>%
  summarise(Obs =n(),
            unemp_yrs=round(mean(unemployed, na.rm = TRUE),3),
            Std.dev. = round(sd(unemployed, na.rm =TRUE),3),
            min = min(unemployed, na.rm = TRUE),
            max = max(unemployed,na.rm = TRUE)
  )

## # A tibble: 4 × 6
##    year    Obs unemp_yrs Std.dev.   min   max
##   <int>  <int>     <dbl>    <dbl> <dbl> <dbl>
## 1  2020 863507     0.072    0.259     0     1
## 2  2021 432672     0.05     0.218     0     1
## 3  2022 411969     0.034    0.182     0     1
## 4  2023 233241     0.035    0.183     0     1

Visualizing unemployment accross years(2020-2023)

Cps20_23%>%
  group_by(year,month)%>%
  summarise(unemp_yrs=mean(unemployed, na.rm = TRUE))%>%
  pivot_wider(names_from = year, 
              values_from = unemp_yrs)%>%
  rename(yr_2020="2020", yr_2021="2021", yr_2022="2022", yr_2023="2023")%>%
  ggplot()+
  geom_line(aes(month, yr_2020, group=1, color="2020"))+
  geom_line(aes(month, yr_2021, group=1, color="2021"))+
  geom_line(aes(month, yr_2022, group=1, color="2022"))+
  geom_line(aes(month, yr_2023, group=1, color="2023"))+
  labs(
    title = "Unemployment rates",
    y="Unemployment_rates", x="Months",
    color = "Years"
  )+
  theme(axis.text.x = element_text(angle = 45, hjust = 0.1, vjust = 0))

## `summarise()` has grouped output by 'year'. You can override using the
## `.groups` argument.

## Warning: Removed 5 rows containing missing values or values outside the scale range
## (`geom_line()`).

#this is the end! thank you.

Regression Analysis (Econ 0211), Lab1 : Introduction

Noe Vyizigiro

March, 2024