variant_signal: Variant Signal

In pgrugwiro/variant: Quantifying Dataframe Changes

Variant Signal

Description

The variant_signal() is the last and most important function in the package. It pipelines all the helper functions and uses their output to compute the signal. It takes as input two data frames and a desired level of significance. It then checks for the data frames validity using the df_check() function and computes the correlation coefficient matrices for both data frames, if they are valid, using the cormat() function. The correlation coefficients are transformed into their z-scores using the fisher_trans() function. Then confidence interval parameters for the mean correlation coefficient of each variable are calculated using the cil() function, overlapping percentages and presence or absence of overlap is determined with the percent_overlap(), no_overlap() functions respectively. The variant_signal() function then calculates average change for the non-overlapping category and compares it to the overall change. This comparison is displayed in terms of percentage and is called “signal”. Additional outputs of the variant_signal() functions are a list of percentage overlaps as computed by the percent_overlap() function, indices for which there is no overlap as determined by the no_overlap() function as well as a confidence interval plot for all the variables and for the non-overlapping variables.

Usage

variant_signal(df1, df2, alpha)

Arguments

df1	dataframe 1
df2	dataframe 2
alpha	level of significance e.g. 0.05

Value

list of columns that show significant change

Examples

## Not run: df1 <- data.frame(matrix(rnorm(100), ncol = 5))
df2 <- data.frame(matrix(rnorm(100), ncol = 5))
variant_signal(df1, df2, 0.05)
## End(Not run)

pgrugwiro/variant documentation built on Aug. 2, 2022, 12:08 p.m.