Calculation of derivatives using the GOLD method

Description Usage Arguments Value Examples

View source: R/doremi.R

Description estimates the derivatives of a variable using the Generalized Orthogonal Local Derivative (GOLD) method described in Deboeck (2010). This method allows calculating over a number of measurement points (called the embedding number) the first derivative with errors uncorrelated with the signal. It was generalized for non-equidistant time points (variable time steps), in order to account for missing observations.


1, time, embedding = 2)



is a vector containing the data from which the derivative is estimated.


is a vector containing the time values corresponding to the signal. Arguments signal and time must have the same length.


is an integer indicating the number of points to consider for derivative calculation. Embedding must be greater than 1 because at least two points are needed for the calculation of the first derivative and at least 3 for the calculation of the second derivative.


Returns a list containing three columns:

dtime- contains the time values in which the derivative was calculated. That is, the moving average of the input time over embedding points.

dsignal- is a data.frame containing three columns and the same number of rows as the signal. The first column is the moving average of the signal over embedding points, the second is the first derivative, and the third is the second derivative.

embedding- contains the number of points used for the derivative calculation, which is constant.


#In the following example the derivatives for the function y(t) = t^2 are calculated.
#The expected results are:
#y'(t) = 2t and y''(t) = 2
time <- c(1:500)/100
signal <- time^2
result <- = signal, time = time, embedding = 5)

doremi documentation built on Aug. 30, 2018, 5:04 p.m.