track: Tracking and Forecasting Interactions in Real Time

In langendorfr/DMMS: Tracking and Forecasting Interactions in Real Time

Description

Calculates sequential Jacobians on time series data using the S-map method developed and detailed in the 2016 Proceedings of the Royal Society B paper titled Tracking and Forecasting Ecological Interactions in Real Time by Deyle, May, Munch, and Sugihara.

Usage

track(time_series, target, dates, theta = 8, manual_block = FALSE,
  causal_probabilities, causal_iterations = 100)

Arguments

time_series	A data frame where each column is a time series.
target	Which time series (column) to calculate the sequential interactions of.
dates	Optional vector of dates.
theta	Parameter to tune the relationship between distance and weight in the linear model. Defaults to 8 which was used by Deyle, May, Munch, and Sugihara.
manual_block	Defaults to FALSE. If your data is stacked spatial replicates, where you have already made the first column for time 1+1 and all others for time t.
causal_probabilities	Causal probabilities, or 1 - p-values from a test for causality, for each interactor (column in the data).
causal_iterations	Defaults to 100. Determines how many times the causal filter is applied. For large datasets this can cause slowdowns which is why the default is only 100.

Value

A data frame where each column is a row of the sequential Jacobian.

References

Deyle, E. R., May, R. M., Munch, S. B., & Sugihara, G. (2016, January). Tracking and forecasting ecosystem interactions in real time. In Proc. R. Soc. B (Vol. 283, No. 1822, p. 20152258). The Royal Society.

Examples

data <- matrix(runif(300, 0, 1), nrow = 100, ncol=3))

track(data, target = 1)

langendorfr/DMMS documentation built on May 14, 2019, 2:55 p.m.