One of the most useful packages I have come across this semester is the stargazer package. It provides a way to create publication quality tables, and a way for researchers to avoid creating new tables each time they tweak their datasets. This package saves users time, and has been welcomed by the R community. It outputs tables in multiple formats; from .txt to LaTex code as well as .html. This tutorial will go through the .txt and .html formats and provide the basic understanding needed to create Summary Statistics Tables and Regression Tables.
This blog will demonstrate some of the applications of this wonderful package
library(stargazer)We will be using the iris data to show the following features of the package.
Summary Statistics
This is similar to the base r function but shows the statistics in a table format.
stargazer(iris, type='text', title=' Iris : Summary Statistics', out='table1.txt')##
## Iris : Summary Statistics
## =============================================================
## Statistic N Mean St. Dev. Min Pctl(25) Pctl(75) Max
## -------------------------------------------------------------
## Sepal.Length 150 5.843 0.828 4.300 5.100 6.400 7.900
## Sepal.Width 150 3.057 0.436 2.000 2.800 3.300 4.400
## Petal.Length 150 3.758 1.765 1.000 1.600 5.100 6.900
## Petal.Width 150 1.199 0.762 0.100 0.300 1.800 2.500
## -------------------------------------------------------------You can also flip the table
stargazer(iris, type='text', title=' Iris : Summary Statistics', out='table1.txt', flip=TRUE)##
## Iris : Summary Statistics
## ===========================================================
## Statistic Sepal.Length Sepal.Width Petal.Length Petal.Width
## -----------------------------------------------------------
## N 150 150 150 150
## Mean 5.843 3.057 3.758 1.199
## St. Dev. 0.828 0.436 1.765 0.762
## Min 4.300 2.000 1.000 0.100
## Pctl(25) 5.100 2.800 1.600 0.300
## Pctl(75) 6.400 3.300 5.100 1.800
## Max 7.900 4.400 6.900 2.500
## -----------------------------------------------------------Regression
I think the most important application of this package is being able to display regression statistics to compare the results. To demonstrate this, I will run three regression models first.
data<-iris
m1<-lm(Sepal.Length~., data)
m2<-lm(Sepal.Length ~ Sepal.Width, data)
m3<-lm(Sepal.Length~ Sepal.Width + Petal.Length, data)Now let us compare the outputs
stargazer(m1, m2, m3, type='text', title='Regression results', out='table2.text')##
## Regression results
## =========================================================================================
## Dependent variable:
## ---------------------------------------------------------------------
## Sepal.Length
## (1) (2) (3)
## -----------------------------------------------------------------------------------------
## Sepal.Width 0.496*** -0.223 0.596***
## (0.086) (0.155) (0.069)
##
## Petal.Length 0.829*** 0.472***
## (0.069) (0.017)
##
## Petal.Width -0.315**
## (0.151)
##
## Speciesversicolor -0.724***
## (0.240)
##
## Speciesvirginica -1.023***
## (0.334)
##
## Constant 2.171*** 6.526*** 2.249***
## (0.280) (0.479) (0.248)
##
## -----------------------------------------------------------------------------------------
## Observations 150 150 150
## R2 0.867 0.014 0.840
## Adjusted R2 0.863 0.007 0.838
## Residual Std. Error 0.307 (df = 144) 0.825 (df = 148) 0.333 (df = 147)
## F Statistic 188.251*** (df = 5; 144) 2.074 (df = 1; 148) 386.386*** (df = 2; 147)
## =========================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01We can easily compare the Rsquare, errors, p values, etc side by side to compare which model fits our data best.