prediction: Predictions for parameterized models on new variable states

In lkshrsch/gLVInterNetworks: Inference of interaction networks based on generalised Lotka Volterra dynamics

Description

Given a fitted model, predict future timepoints starting from given starting points or random starting points. Compare prediction with test data from the inference result object. Only for result assessment.

Usage

prediction(result, data)

Arguments

result
data

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (result, data) 
{
    x0 <- data$testData[1, -1]
    pred <- solveLV_bio(result$Parms, data$testData[1:nrow(data$testData), 
        1], x0)
    Coef_of_determination <- R_squared(pred, data$testData)
    result$R_squared_test_Data <- Coef_of_determination
    r <- c()
    for (i in 1:10) {
        x0 <- runif(n = data$species, 0, 1)
        new <- solveLV_bio(data$Parms, times = 0:50, States = x0)
        pred2 <- solveLV_bio(result$Parms, new[, 1], x0)
        r[i] <- R_squared(pred2, new)
    }
    result$R_squared_new_Data <- mean(r)
    return(result)
  }

lkshrsch/gLVInterNetworks documentation built on May 21, 2019, 7:33 a.m.