umxSummary: report algebra variables

[tbates]

Currently neither OpenMx::summary nor umxSummary show users the values of any algebras in a RAM model.

Very useful to add this as default output, with estimate, SE and P(>|Z|): if they add them, they probably want to see the value :-)

library(mlbench)
data(BostonHousing2)

BostonHousing2$log_crim  = log2(BostonHousing2$crim)
BostonHousing2$nox  = BostonHousing2$nox*100

m1 = umxRAM(data = BostonHousing2,
	"#crime_model
	cmedv ~ log_crim + b1*nox; 
	nox   ~ a1*rad + a2*log_crim
	i_1 := a1*b1
	i_2 := a2*b1"
)

Outputs the parameters and fit, but not the i_1 and i_2 algebras:

Table: Model Parameter loadings

name	Estimate	SE
b1	-0.144	0.051
a1	-0.302	0.068
log_crim_to_cmedv	-0.925	0.188
a2	3.648	0.191
cmedv_with_cmedv	65.499	4.118
nox_with_nox	48.798	3.068
rad_with_rad	75.667	4.757
log_crim_with_rad	23.132	1.584
log_crim_with_log_crim	9.710	0.610
one_to_cmedv	29.470	2.995
one_to_nox	62.457	0.900
one_to_rad	9.549	0.387
one_to_log_crim	-1.126	0.139

Model Fit: χ²(1) = 0.08, p = 0.777; CFI = 1.001; TLI = 1.004; RMSEA = 0

[tbates]

Probably involves ?mxSE, so

if (length(model@algebras)>0){
    for(thisAlg in names(m3@algebras)){
       # OpenMx algebras needed be values, so just emit what we find, more free-form
       b  = mxEvalByName(thisAlg, model)
       SE = mxSE(thisAlg, model, forceName = TRUE)
       p = 2*pnorm(abs(b/SE), lower.tail = FALSE)
       print(paste0(thisAlg, umxAPA(b, SE), SE, p))
   }

       # And just emit larger matrices more free-form? or break into a list of cells?      
   }
}

[tbates]

umxSummary now reports algebra values

Algebra'i_1' = 0.043CI95[0.008, 0.079]. p-value = 0.017
Algebra'i_2' = -0.525CI95[-0.89, -0.16]. p-value = 0.005