Ever wanted what’s the best way to learn a new language ? well my secret is in using things I already love like R for Example . I make comparisons to familiar languages that I start to learn . It works like a charm all the time

We are going to work a lot with the dplyrpackage. You need to load the package just once when you open your environment. If you have any question, please check our other material to start learning R. However, the best way to learn is try different approaches (;

Working directory

/*Working directory path*/
pwd 

/*Changes the working directory path*/
cd c:\myfolder\data

#Working directory path
getwd()

#Changes the working directory path
setwd("my_path")

Packages

/*Install package*/
ssc install abc

Once is installed, we don’t need to “load it”

#Working directory path
install.packages("tidyverse")

#Load library
library(tidyverse)

Help

/*Get help with the command regression*/
help regress

#Help with mutate function from dplyr package
?mutate()

Load Data

CSV Files

/*Import csv file*/
import delimited "my_data.csv", clear

library(readr)

#Use read_csv2
df_data <- read_csv2("data.csv")

#Using read.table function with comma separator
df_data <- read.table("my_data.csv", 
                      sep = ",", 
                      header = TRUE)

Excel Files

/*Import excel file*/
import excel "df_data.xlsx", 
sheet("Sheet1") firstrow clear

#With readxl package
library(readxl)
df_data <- read_excel("my_data.xlsx", 
                      sheet = "Sheet1")

#With openxlsx package
library(openxlsx)
df_data <- read.xlsx("my_data.xlsx", 
                     sheet = "Sheet1")

Stata Files (.dta)

/*Import excel file*/
use "df_data.dta", clear

#Load the package
library(haven)            

df_data <- read_dta("df_data.dta")

Explore Data

Remember the difference between R and Stata. While in R you need to specify which object (or data) you want to work with, in Stata, the variables are loaded, so it’s easier to work with them.

/* Provides the structure of the dataset */
describe

/*Basic descriptive statistics */
summarize variable1 variable2

/*Lists the variables in the dataset */
ds

/* First 10 observations */
list in 1/10 

/* Last 10 observations */
list in -1/10

/* Show first 10 observations of the */
/*  first three variables of our data */
list var1-var3 in 1/10






/*View data */
browse

#Provides the structure of our data 
str(df_data)

#Basic descriptive statistics
summary(df_data)

#Lists the variables of our data 
names(df_data)

#Show first 10 observations of our data 
head(df_data, n = 10)

#Show last 10 observations of our data 
tail(df_data, n = 10)

#Show first 10 observations of the 
# first three variables of our data 
df_data[1:10, 1:3]

# With dplyr, this would be other option
library(dplyr)
df_data %>% slice(1:10) %>% select(1:3)

#View data
View(df_data)

Missing Data

/* Provides the structure of the dataset */
missing variable1 variable2 variable3

#Check missing data for our entire database
is.na(df_data)

#Check missing data for our entire database
is.na(df_data$variable1)

Rename variables

/* How to rename */
rename oldname newname

rename lastname lastname2
rename firstname firstname2
rename studentstatus studentstatus2
rename averagescoregrade avgscore

#Rename the variables of our data
names(df_data) <- c("newvar1", "newvar2")

#With the dplyr package

library(dplyr)
df_data_final <- rename(df_data, 
                        newvar1 = var1, 
                        newvar2 = var2)

Label variables

label variable w "Weight"
label variable y "Output"
label variable x1 "Predictor 1"
label variable x2 "Predictor 2"
label variable age "Age"
label variable sex "Gender"

#Load labelled package
library(labelled)

#Define the value labels for the codes
labels <- c("Option 1", "Option 2", 
            "Option 3", "Option 4", 
            "Option 5")
df_data$code <- labelled(df_data$code, 
                         labels = labels)

Value labels

/* Value labels for the codes */
label define label1 1 "Option 1" 
2 "Option 2" 3 "Option 3" 4 "Option 4" 
5 "Option 5"

/* Assign the value labels to the codes */
label values code label1

#Load labelled package
library(labelled)

#Define the value labels for the codes
labels <- c("Option 1", "Option 2", 
            "Option 3", "Option 4", 
            "Option 5")
df_data$code <- labelled(df_data$code, 
                         labels = labels)


#Define the value labels for the codes
labels <- c("Option 1", "Option 2", 
            "Option 3", "Option 4", 
            "Option 5")
df_data$code <- factor(df_data$code, 
                       levels = 1:5, 
                       labels = labels)

In this approach, you can assign values to numeric or character variables

Group variables

/* Value labels for the codes */
collapse (mean) var1, by(var2, var3)

list var2 var3 var1

#Load dplyr package
library(dplyr)

#We group the data by variables 2 and 3, 
#calculate the mean of variable 1
df_data_group <- df_data %>%
  group_by(var2, var3) %>%
  summarise(mean_var1 = mean(var1))

Merge two datasets

/* Exact match on the key variable(s). */
use dataset1.dta, clear
merge 1:1 id using dataset2.dta
/* Each observation in the first dataset */
merge 1:m id using dataset2.dta
/* Each observation in the second dataset */
merge m:1 id using dataset2.dta
/* Each observation many-to-many merge */
merge m:1 id using dataset2.dta

Remember that we need to have our data1 in the same path

#Load dplyr package
library(dplyr)

#1:1 using inner join
df_merged_inner <- inner_join(df_1, df_1, 
                              by = "id1")

#1:m using left join
df_merged_left <- left_join(df_1, df_1, 
                            by = "id1")

#m:1 using right join
df_merged_right <- right_join(df_1, df_1, 
                              by = "id1")

#m:m using full join
df_merged_full <- full_join(df_1, df_1, 
                            by = c("id1", 
                                   "id2"))
#You can merge with more tha one variable

Create sequence/ids

/* Sequence of numbers from 1 to n */
gen seq = _n

/* Sequence with the total number of 
observations */

gen total_seq = _N

v_seq <- seq(from = 1, to = 10, by = 1)

#Load dplyr package
library(dplyr)

df_final <- df %>%
  mutate(id = 1:n())

Drop variables

/* Drop variable */
drop var1 var2

/* Drop all variable and keep */
drop _all
keep var1 var2 var3

#Load dplyr package
library(dplyr)

#Drop some variables
df_new_data <- df_old %>% select(-var1, -var2)

#Keep specific variables
df_new_data <- df_old %>% select(var1, var2)

Frequencies/Tabulation

/* Cross-tabulate 3 variables*/
tabulate var1 var2 var3

/* Cross-tabulate with row and 
column percentages */
tabulate var1 var2, row col

#Cross-tabulate 3 variables
table(var1, var2, var3)

# The ~ symbol separates 
# the row and column variables
xtabs(~var1+var2, data = dataset)

Statistics

Summary

/* Summary of variables of interest*/
summarize var1 var2 var3

#Summary for entire data
summary(dataset)

#Summary for a variable
summary(dataset$variable)

Linear Model

/* Fit a linear model */
regress mpg weight length foreign

/* Print summary */
estimates table

#Fit a linear model
model_lm <- lm(mpg ~ wt + qsec + vs, 
               data = df_my_data)

#Print summary
summary(model_lm)

Logistic Model

/* Fit a logistic regression model */
logit foreign weight length

/* Print summary */
estimates table

#Fit a logistic regression model
model_log <- glm(vs ~ wt + qsec, 
                 data = df_my_data, 
                 family = binomial())

#Print summary
summary(model)

Poisson Model

/* Fit a Poisson regression model */
poisson accidents weight length foreign

/* Print summary */
estimates table

#Fit a Poisson regression model
model_poiss <- glm(am ~ wt + hp, 
                   data = df_my_data, 
                   family = poisson())

#Print summary
summary(model)

Plots

Scatter plot

/* Create scatter plot */
scatter price mpg

# Load data and ggplot2
library(ggplot2)
data(df_my_data)

#Create scatter plot
ggplot(df_my_data, aes(x = mpg, y = wt)) + 
  geom_point() + #Here is the scatter
  xlab("Miles per gallon") +
  ylab("Weight")

Line plot

/* Create Line plot */
twoway line le_w le_y, xlabel(1950(10)2010)

# Load data and ggplot2
library(ggplot2)
data(df_my_data)

#Create line plot
ggplot(df_my_data, aes(x = year, y = pop)) + 
  geom_line() + #Here is the line
  xlab("Year") +
  ylab("Population")

Bar plot

/* Create a Bar plot */
graph bar (count) rep78, over(foreign)

# Load data and ggplot2
library(ggplot2)
data(df_my_data)

#Create bar chart
ggplot(data = df_my_data, 
       aes(x = factor(cyl))) +
  geom_bar() + #Here is the bar
  xlab("Number of cylinders") +
  ylab("Count")

I also want to provide a good cheat sheet that contains more information about the differences between Stata and R.

Cheat Sheet: Stata to R

RStata

Believe it or not, R has a package called RStata. This package Stata interface allows the user to execute Stata commands (both inline and from a .do file) from R. I leave herethe package. Looks interesting; however, I personally prefer to use a language as expected.

Conclusion

In conclusion, the decision to transition from one to the other should be made based on the user’s specific needs and preferences. This notebook shows you the differences between Stata and R. At the end, use the language you prefer, but in my own experience, R was a life changer.

References

Note

Jutkowitz, E., Pizzi, L. T., Shewmaker, P., Alarid‐Escudero, F., Epstein‐Lubow, G., Prioli, K. M., … & Gitlin, L. N. (2023). Cost effectiveness of non‐drug interventions that reduce nursing home admissions for people living with dementia. Alzheimer’s & Dementia.