Forum: developers

RE: coefTable/sampleSelection [ Reply ] By: Ott Toomet on 2009-05-10 15:30	[forum:1526]
OK. Lets go for that. Will you start it yourself?

RE: coefTable/sampleSelection [ Reply ]
By: Arne Henningsen on 2009-05-09 18:54

[forum:1524]

Yes, creating larger (and still rather homogenous) packages would be great. However, this seems to be infeasible given the huge number of (heterogenous) package developers. The purpose of the utilMisc (or similar) package would be to keep functions until they are moved into a more suitable package (e.g. base or stats) or they could be in this package forever. I suggest that this package should not have any dependencies. It should include (generic) functions and methods that are used by more than one package. Hence, these packages depend on the utilMisc (or similar) package but do not depend on each other, which reduces (recursive) dependencies. What do you think?

RE: coefTable/sampleSelection [ Reply ] By: Ott Toomet on 2009-05-09 13:00	[forum:1521]
This is probably the best we can do currently. What should be the scope/purpose of the package? The problem is just that the number of packages at CRAN is approaching 2000 with a long incomprehensible (for humans) dependence relationship between them. I'd like to see larger and more comprehensive packages in the long run. However, I don't know if it is feasible.

RE: coefTable/sampleSelection [ Reply ]
By: Arne Henningsen on 2009-05-09 08:02

[forum:1518]

Yes. The sampleSelection package is not a place for "coefTable". Likewise, the "micEcon" package is not a good place for (at least some of) the utility functions "colMedians", "compPlot", "insertCol", "insertRow", "rSquared", "semidefiniteness", "symMatrix", "triang", "vecli", "vecli2m", "veclipos". Since these functions are rather utility functions than econometric functions, I suggest that we should create a new package for miscellaneous utility functions, e.g. "utilMisc" or so. What do you think?

coefTable/sampleSelection [ Reply ] By: Ott Toomet on 2009-05-03 19:05	[forum:1468]
sampleSelection includes a number of utilities which are not directly related to selection models. What to do with e.g. coefTable, a handy utility to print coefficients? Currently I have marked sampleSelection as a required package but this seems not too good idea...