Automated Visualization of Statistical Tests.
visStatistics
Visualization of the statistical hypothesis test between two groups of categorical or numerical data.
The function visstat()visualizes the statistical hypothesis testing between two groups of data, where varsample is the dependent variable (or response) and varfactor is the independent variable (feature). The statistical hypothesis test with the highest statistical power and fulfilling the assumptions of the corresponding test is performed and visualized. A graph displaying the raw data accordingly to the chosen test as well as the test statistics is generated. Furthermore visstat() returns the corresponding test statistics as text. The automated workflow is especially suited for browser based interfaces to server-based deployments of R. Implemented tests: lm(), t.test(), wilcox.test(), aov(), oneway.test(),kruskal.test(), fisher.test(),chisqu.test().
Installation from GitHub
- Firstly, you need to install the devtools package. You can do this from CRAN. Invoke R and then type
install.packages("devtools") - Load the devtools package.
library(devtools) - Install the package from the github- repository
install_github("shhschilling/visStatistics") - Load the package
library(visStatistics) - Help on the function usage
?visstat
Examples
Trees data set: Linear regression
visstat(trees,"Girth","Height")
Iris data set: Kruskal-Wallis test
visstat(iris,"Petal.Width", "Species")
InsectSprays data set: ANOVA
visstat(InsectSprays,"count","spray")
InsectSprays data set: Welch two sample t.test
InsectSpraysAB <- InsectSprays[ which(InsectSprays$spray == 'A'| InsectSprays$spray == 'B'), ] #select only sprays 'A und 'B'
InsectSpraysAB$spray = factor(InsectSpraysAB$spray)
visstat(InsectSpraysAB,"count","spray")
ToothGrowth data set: Wilcoxon rank sum test with continuity correction
visstat(ToothGrowth,"len", "supp")
HairEyeColor data set: Pearson's Chi-squared test
HairEyeColorMale = counts_to_cases(as.data.frame(HairEyeColor[,,1]))
visstat(HairEyeColorMale,"Hair","Eye")