/
NEWS
29 lines (24 loc) · 1.54 KB
/
NEWS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Changes in version 0.3.5
o lvnet is now no longer supported, with most functionality moved to the psychonetrics package. A message now warns users of this on loading the package.
Changes in Version 0.3.3
o Added 'refitAll' argument to lvnetLasso to refit all models in the LASSO path before evaluating EBIC. This leads to similar functionality as qgraph::ggmModSelect(.., stepwise = FALSE), although the performance is slower.
o The argument 'refit' has been renamed 'refitFinal'
o Removed a warning on depracated dplyr functions
o Fixed a bug in counting the number of parameters in RNM LASSO estimation
Changes in Version 0.3.3
o 'ggmFit' is no longer exported by lvnet
Changes in Version 0.3.2
o Summary output now contains parameter labels
o Equality constraints now supported by supplying labels in model matrices
Changes in Version 0.3.1:
o 'ggmFit' is now moved to qgraph and exported by lvnet
o Several updates to ggmFit in qgraph
o Added 'lvnetRefit' to refit an lvnet model to new data
Changes in Version 0.3:
o 'scale' now defaults to FALSE!
o Due to this change, some values such as the BIC will be different
o Fixed a bug where higher order latent variables were not usuable
o Added the 'mimic' argument. mimic = "lavaan" will mimic lavaan results (now default).
o lvnet should now return the same fit as lavaan when std.lv=TRUE is used in lavaan.
o Added the function 'lav2lvnet' to easily generate lvnet input from a lavaan model
o Added 'ggmFit' to fit a given partial correlation network (Gaussian graphical model; GGM)