find_difference: Return the regions in which the smooth is significantly...

In itsadug: Interpreting Time Series and Autocorrelated Data Using GAMMs

Return the regions in which the smooth is significantly different from zero.

Description

Return the regions in which the smooth is significantly different from zero.

Usage

find_difference(mean, se, xVals = NULL, f = 1, as.vector = FALSE)

Arguments

mean	A vector with smooth predictions.
se	A vector with the standard error on the smooth predictions.
xVals	Optional vector with x values for the smooth. When xVals is provided, the regions are returned in terms of x- values, otherwise as indices.
f	A number to multiply the se with, to convert the se into confidence intervals. Use 1.96 for 95% CI and 2.58 for 99%CI.
as.vector	Logical: whether or not to return the data points as vector, or not. Default is FALSE, and a list with start and end points will be returned.

Value

The function returns a list with start points of each region (start) and end points of each region (end). The logical xVals indicates whether the returned values are on the x-scale (TRUE) or indices (FALSE).

Author(s)

Jacolien van Rij

See Also

Other Utility functions: convertNonAlphanumeric(), corfit(), diff_terms(), missing_est(), modeledf(), observations(), print_summary(), refLevels(), res_df(), summary_data(), timeBins()

Examples

data(simdat)

# Use aggregate to calculate mean and standard deviation per timestamp:
avg <- aggregate(simdat$Y, by=list(Time=simdat$Time),
    function(x){c(mean=mean(x), sd=sd(x))})
head(avg)
# Note that column x has two values in each row:
head(avg$x)
head(avg$x[,1])

# Plot line and standard deviation:
emptyPlot(range(avg$Time), c(-20,20), h0=0)
plot_error(avg$Time, avg$x[,'mean'], avg$x[,'sd'], 
   shade=TRUE, lty=3, lwd=3)

# Show difference with 0:
x <- find_difference(avg$x[,'mean'], avg$x[,'sd'], xVals=avg$Time)
# Add arrows:
abline(v=c(x$start, x$end), lty=3, col='red')
arrows(x0=x$start, x1=x$end, y0=-5, y1=-5, code=3, length=.1, col='red')

itsadug documentation built on June 17, 2022, 5:05 p.m.