taylor | R Documentation |
The power law relationship between variance and mean is known as Taylor's law, which is defined as: var(Y) = a*mean(Y)^b. We can obtain the power as the slope in log-scale: log(var(Y)) = log(a)+b*log(mean(Y))
taylor(x, type = "taylor", boot = 0, interval = "prediction", lower.conf = 0.025, upper.conf = 0.975, pseudo = 0, col = "black", header = "", label = FALSE, plot = TRUE)
x |
a matrix, mean and variance are computed row-wise |
type |
the type of plot to do: mean.var (mean vs variance), boxplot (row-wise), taylor (powerlaw fitted to mean vs variance) |
boot |
compute confidence interval for individual data points through bootstrapping with given number of iterations (only for type=taylor) |
interval |
compute prediction or confidence interval (for type=taylor) |
lower.conf |
lower limit of confidence interval for both regression line and individual data points (for type=taylor) |
upper.conf |
upper limit of confidence interval for both regression line and individual data points (for type=taylor) |
pseudo |
add a pseudo count to deal with zeros in log-log plot (for type=taylor) |
col |
the color of the dots |
header |
header string |
label |
label points interactively (only for type=taylor, escape to stop labeling) |
plot |
whether to display the plot |
For plot type taylor, confidence intervals for individual data points are estimated by first resampling columns with replacement from the input matrix and then recomputing log means and log variances on the bootstrapped data. The confidence interval shown for the regression line is the prediction interval.
for type taylor, the slope, p-value, adjusted R2 of the Taylor law as well as the log means and variances and for boot>0, the lower and upper values of log mean and variance bootstraps are returned (slope, pval, adjR2, logmeans, logvars, lowerConfMean, upperConfMean, lowerConfVar, upperConfVar)
L.R. Taylor (1961). Aggregation, variance and the mean. Nature 189, 732-735.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.