Bivariate Automatic Analysis.

BiVariAn

Overview

The purpose of ‘BiVariAn’ is to facilitate bivariate analysis by integrating functions that allow the results of such analyses to be represented graphically and in text.

Currently the package can generate summary tables of analysis data for: Analysis of continuous variables across two groups using Student’s t-tests (OR Welch’s t-tests according to homogeneity of variances), Mann-Whitney U; analysis of dichotomous variables using Chi-square and Fisher’s exact test; analysis of normality of raw data using the Shapiro-Wilk test.

It is also possible to generate graphs automatically with or without labels on the variables with the possibility of choosing a theme from the ‘ggplot2’ package.

Currently under development is the option to generate a generic interpretation of the statistical tests, temporarily available only in Spanish, but there are plans to include more languages.

In addition to bivariate analyses, regression analyses with predictor selection using the stepwise backward method for linear, logistic and Firth penalized logistic regression models can be performed in an automated manner.

Note: The use of the package does not replace individual customized analysis, but rather streamlines the process of producing the results. We encourage users to perform statistical analyses based on the recommendations for each type of study.

Installation

You can install the development version of BiVariAn from GitHub with:

# install.packages("pak")
pak::pak("AndresFloresG/BiVariAn")

Or you can install the official CRAN version with:

install.packages("BiVariAn")

Example

Loading the package

library(BiVariAn)

Render an automatic Shapiro-Wilk’s table of a simple dataset

auto_shapiro_raw(cars)

shapiro.test(cars$speed)
#> 
#>  Shapiro-Wilk normality test
#> 
#> data:  cars$speed
#> W = 0.97765, p-value = 0.4576

Return Shapiro-Wilk’s results as a dataframe

auto_shapiro_raw(cars, flextableformat = FALSE)
#>       Variable p_shapiro  Normality
#> speed    speed   0.45763     Normal
#> dist      dist    0.0391 Non-normal

Render an automatic Shapiro-Wilk’s table of a more complex dataset

# Load riskCommunicator to access Framingham dataset
library(riskCommunicator)

# Load dplyr to select specific columns
library(dplyr)
data(cvdd)

For shapiro.test, sample size must be between 3 and 5000

Let’s select only 300 observations (arbitrary)

set.seed(081224)
ex_sample<-slice_sample(cvdd, n=300)

Now, let’s select specific columns from the database

auto_shapiro_raw(ex_sample %>% select(TOTCHOL, SYSBP, DIABP, BMI, HEARTRTE))

Common use of shapiro.test

shapiro.test(ex_sample$TOTCHOL)
#> 
#>  Shapiro-Wilk normality test
#> 
#> data:  ex_sample$TOTCHOL
#> W = 0.98654, p-value = 0.007891

Return the same Shapiro-Wilk’s results as a dataframe

auto_shapiro_raw(ex_sample %>% select(TOTCHOL, SYSBP, DIABP, BMI, HEARTRTE), flextableformat = FALSE)
#>          Variable p_shapiro  Normality
#> TOTCHOL   TOTCHOL   0.00789 Non-normal
#> SYSBP       SYSBP   <0.001* Non-normal
#> DIABP       DIABP   <0.001* Non-normal
#> BMI           BMI   <0.001* Non-normal
#> HEARTRTE HEARTRTE   <0.001* Non-normal