Computes precision and recall statistics separating observed from hidden variables, and (FPR,FNR) for hidden variables .

Computes precision and recall statistics separating observed from hidden variables, and (FPR,FNR) for hidden variables .

ppvtpr(probs, G, r, thresh = 0.5)

Arguments

probs	Matrix of estimated edges probabilities.
G	Original graph.
r	Number of missing actors.
thresh	Required threshold for criteria computations, default to 0.5.

Value

• PPV: precision of the whole data.

• PPVH: precision regarding observed data.

• PPVO: precision regarding hidden data.

• TPR: recall of the whole data.

• TPRH: recall regarding observed data.

• TPRO: recall regarding hidden data.

• FPRH: false Positive Rate of hidden data.

• FNRH: false Negative Rate of hidden data.

Examples

data=generate_missing_data(n=100,p=10,r=1,type="scale-free", plot=FALSE)
PLNfit<-norm_PLN(data$Y)
MO<-PLNfit$MO
SO<-PLNfit$SO
sigma_O=PLNfit$sigma_O
#-- use true clique for example
initClique=data$TC
#-- initialize the VEM
initList=initVEM(cliqueList=initClique,sigma_O, MO,r=1 )
nestorFit=nestorFit(MO,SO, initList=initList, maxIter=3,verbatim=1 )
#> 
#> nestor ran in 0.212secs and 3 iterations.
#-- obtain criteria
ppvtpr(nestorFit$Pg,r=1, data$G)
#> $PPV
#> [1] 1
#> 
#> $PPVH
#> [1] 1
#> 
#> $PPVO
#> [1] 1
#> 
#> $TPR
#> [1] 1
#> 
#> $TPRH
#> [1] 1
#> 
#> $TPRO
#> [1] 1
#> 
#> $FPRH
#> [1] 0
#> 
#> $FNRH
#> [1] 0
#>