permuspliner: Permuspliner

In RRShieldsCutler/splinectomeR: Test for significance in longitudinal data with one or two populations

Permuspliner

Description

Tests for a significant difference between two groups overall.

Usage

Permutation to test whether there is a non-zero trend among a set
of individuals/samples over a continuous variable -- such as time. So, there
does not need to be two groups in this test. The x variable datapoints are
permuated within each case/individual, thus maintaining the distribution in
the y component but shuffling the hypothesized trend.

Arguments

data	A dataframe object containing your data.
xvar	The independent variable; is continuous, e.g. time.
yvar	The dependent variable; is continuous, e.g. temperature.
category	The column name of the category to be tested.
cases	The column name defining the individual cases, e.g. patients.
groups	If more than two groups, the two groups to compare as character vector.
perms	The number of permutations to generate
retain_perm	Retain permuted spline data for permutation confidence interval plotting (set to FALSE for less memory)
test_direction	Test whether the groups are significantly 'more' or significantly 'less' distinct than expected by random chance. Default is 'more'.
set_spar	Set the spar parameter for splines
cut_low	Remove individual cases with fewer than _ observations
ints	Number of x intervals over which to measure area
quiet	Silence all text outputs
set_tol	In rare cases, must manually set the tol parameter (default 1e-4)
cut_sparse	Minimum number of total observations necessary per group to fit a spline (default 4)

Examples

result <- permuspliner(data = ChickWeight, xvar = 'Time',
             yvar = 'weight', category = 'Diet',
             cases = 'Chick', groups = c(1,2))
result$pval

RRShieldsCutler/splinectomeR documentation built on April 24, 2022, 2:20 a.m.