assessment: Assess results from inference algorithm

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

Description

Assesses inference result comparing estimated parameters with original simulated parameters (thus only works for in silico data), and stores the comparison in three objects: Quantitative, Qualitative1 and Qualitative2

Usage

assessment(result, Parms)

Arguments

result
Parms

Value

Quantitative	Ratio of parameter estimates that contain the original parameter within their computed .95 confidence interval
Qualitative1	Ratio of edges in the inferred network that were correctly recovered
Qualitative2	Ratio of significant edges in the inferred network that were correctly recovered

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, Parms) 
{
    "still need to check how many zeros are in Parms and correct for qualitative..."
    quantitative <- correctly_intervalized_estimates(c(result$Parms), 
        c(result$SE), c(Parms))
    quantitative <- sum(quantitative)/length(quantitative)
    qualitative <- compare_matrix(c(result$Parms), c(Parms))
    qualitative <- sum(qualitative)/length(qualitative)
    is_not_significant <- correctly_intervalized_estimates(c(result$Parms), 
        c(result$SE), rep(0, length(Parms)))
    is_zero <- c(Parms) == 0
    qualitative2 <- is_not_significant == is_zero
    qualitative2 <- sum(qualitative2)/length(qualitative2)
    result$quantitative = quantitative
    result$qualitative1 = qualitative
    result$qualitative2 = qualitative2
    return(result)
  }

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