Embed: Time-delay embedding

In rEDM: Empirical Dynamic Modeling ('EDM')

Time-delay embedding

Description

Takens time-delay embedding of one or more columns, returned as a named data.frame matching pyEDM's Embed().

Usage

Embed(dataFrame = NULL, E = 0, tau = -1, columns = "",
      includeTime = FALSE, pathIn = "./", dataFile = "")

Arguments

dataFrame	A data.frame of input data. Column 1 is the time vector.
E	Embedding dimension.
tau	Embedding delay in time-column rows (negative selects past lags).
columns	Column name(s) to embed: a whitespace-separated string, or a character vector of column names.
includeTime	If TRUE, prepend column 1 of dataFrame (the time vector), keeping its original name, as the first output column.
pathIn	Directory for dataFile (used when reading input from disk).
dataFile	CSV file name to read as input; when set it takes precedence over dataFrame.

Details

Each columns item produces E lagged vectors. For column c and step i = 0..E-1 the lag is i * |tau|, and the column is named c(t-lag) when tau < 0 or c(t+lag) when tau > 0. For example columns X, Y with E = 2, tau = -1 yield X(t-0) X(t-1) Y(t-0) Y(t-1); with tau = -2 they yield X(t-0) X(t-2) Y(t-0) Y(t-2).

If dataFile is supplied the CSV at file.path(pathIn, dataFile) is read (preserving column names) and used as the input, taking precedence over dataFrame.

All rows are retained; the first |tau| * (E-1) partial rows contain NA.

Value

A data.frame of the embedding (with the time column first when includeTime = TRUE).

Examples

Embed(data.frame(x = 1:10), E = 3, tau = -1, columns = "x")

rEDM documentation built on Aug. 30, 2026, 9:07 a.m.