Description Usage Arguments Details Examples
View source: R/Variance_Threshold.R
I was surprised to find that most all of the the work in detecting changes in varance are almost always associated with time-series objects. There are several packages that address change-in-means situations in regression problems. Given this, we will convert the problem to a change-in-mean by calculating a moving window of standard deviation values and then find the change-point on these.
1 | changePoint(data, window.length = 0.01, interval = TRUE)
|
data |
A data frame where each row is a Peptide and column represents a repeated measurement. |
window.length |
The proportion of the data to be used for the sliding window. |
interval |
A TRUE/FALSE flag that indicates if a bootstrap confidence interval should be calculated. |
Once the sliding-window standard deviations are calculated, we use a "mob" regression tree that fits a linear model on each leaf is run and we just fit a simple intercept only model. Thus the leaves will have an intercept and different variance terms.
To calculate a confidence interval, we do a bootstrap procedure where we bootstrap by resampling the peptides, so all repeats of a peptide is removed included as a group.
1 2 3 4 5 6 7 8 9 10 11 | df <- data.frame(x=seq(0,1,by=0.005)) %>%
mutate(y1 = x + rnorm(length(x), sd=ifelse(x<0.5, .1, .3)),
y2 = x + rnorm(length(x), sd=ifelse(x<0.5, .1, .3)),
y3 = x + rnorm(length(x), sd=ifelse(x<0.5, .1, .3)))
ggplot(df, aes(x=x)) +
geom_point(aes(y=y1)) +
geom_point(aes(y=y2)) +
geom_point(aes(y=y3))
changePoint(df[, 2:4], .01)
data = df[,2:4]
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.