aaronwolen / mxtx

mxtx: OpenMx Transcriber
0 stars 0 forks source link

initial feedback on `mxReport` #12

Closed tpyork closed 7 years ago

tpyork commented 7 years ago
  1. OBSERVED DATA SUMMARIES; this reported only once.

    • mean
    • variance
    • min
    • max
    • missing
    • covariance (correlation) matrix; use only outcome variables
    • could add simple histogram plots here
  2. SUMMARIES FOR EACH MODEL FIT; each should have its own link.

    • expected mean
    • expected covariance matrix; use only outcome variables
    • num. observed statistics
    • num. free parameters
    • -2LL
    • DF
    • AIC
    • report estimates of free parameters
  3. SUMMARY FOR mxCompare(); this should be a single table

tpyork commented 7 years ago

12 comment updated

aaronwolen commented 7 years ago

2017-10-24 09 53 40 - 37 54402 -77 4312 - apple iphone 6s

aaronwolen commented 7 years ago

I modified this design slightly to make the results more explicitly mirror the user's input:

Input Sections in produced report
mxReport(AdeFit) Section 1 (AdeFit), Section 2 (AdeFit)
mxReport(list(AdeFit, AeFit) Section 1 (AdeFit), Section 2 (AdeFit), Section 2 (AeFit)
mxReport(AdeFit, c(AdeFit, AeFit)) Section 1 (AdeFit), Section 2 (AdeFit), Section 3 (AdeFit vs AeFit)
mxReport(AdeFit, AeFit, c(AdeFit, AeFit)) Section 1 (AdeFit), Section 2 (AdeFit), Section 2 (AeFit), Section 3 (AdeFit vs AeFit)

So the rules are:

  1. For the first instance of an MxModel produce an Observed Data Summary Section and an Expected Data Summary Section
  2. For all subsequent list elements:
    • if it contains a single MxModel, produce an Expected Data Summary Section
    • it it contains multiple MxModels, produce a Model Comparison Section

Does that make sense?

tpyork commented 7 years ago

This looks a lot nicer ! Not sure where you're at but there are a few things I did not see:

aaronwolen commented 7 years ago

Fixed in 7f1f076b.