runn_mean_pred: Prediction based on a running mean

In chillR: Statistical Methods for Phenology Analysis in Temperate Fruit Trees

Prediction based on a running mean

Description

Function to predict values based on a running mean (or another function) of a numeric vector.

Usage

runn_mean_pred(
  indep,
  dep,
  pred,
  runn_mean = 11,
  na.rm = FALSE,
  exclude_central_value = FALSE,
  FUN = mean
)

Arguments

indep	numeric vector of independent variables, should be sequential
dep	numeric vector of dependent variables
pred	numeric vector of values to be predicted
runn_mean	number of vector elements to use for calculating the running mean
na.rm	ignore NA values when calculating means. Defaults to FALSE.
exclude_central_value	exclude central value in calculating means. Defaults to FALSE.
FUN	function to be applied. For a running mean, this is usually mean (the default), but other functions can also be specified here (the na.rm parameter won't work then, and the function has to be dependent on one numeric variable only.

Details

The running mean calculation that underlies the prediction is based purely on the sequence of observed values, without accounting for any variation in intervals of the independent data. This means that the function performs best with regularly spaced independent variables. Note that the function will return NA when asked to predict values that are outside the range of independent values provided as input. The prediction results are computed by linearly interpolating between the running mean values determined for the nearest neighbors of the value that is to be predicted.

Value

list of two elements, with $x containing the values to be predicted and $predicted the predicted values

Author(s)

Eike Luedeling

Examples

indep<-(1:100)
dep<-sin(indep/20)+rnorm(100)/5
pred<-c(12,13,51,70,90)

predicted<-runn_mean_pred(indep,dep,pred,runn_mean = 25)

plot(dep~indep)
points(predicted$predicted~predicted$x,col="red",pch=15)

chillR documentation built on Nov. 28, 2023, 1:09 a.m.