Debian Codegraph @

BUGS

raw
1 Here is a list of known bugs.
2
3
4 2020-10
5
6 rrvglm() does not do half stepping so one iteration can be worse
7 than the previous. It seems the latest estimate of A and/or C
8 does not translate into an improvement in the next IRLS iteration.
9
10
11
12 2016-06
13
14 lrtest(zipoissonff.object, poissonff.object) fails.
15
16
17
18
19 2016-05
20
21 rcim() with alaplace2() may fail.
22
23
24
25
26 2014-02
27
28 The subset argument of vgam() may not work, especially with
29 multiple responses. To get around this, use subset() to create
30 a smaller data frame and then feed that into vgam().
31
32
33
34
35 2013-11
36
37 vgam() can only handle constraint matrices cmat, say, such that
38 t(cmat) %*% cmat is diagonal.
39
40
41
42
43 2013-07
44
45 quasipoisson()'s scale parameter estimate does not handle
46 prior weights correctly.
47
48
49
50
51 2012-09
52
53
54 loge('a', short = FALSE, inverse = FALSE)
55 loge('a', short = FALSE, inverse = TRUE)
56 give the same answer.
57
58
59
60 Coef(vglm.dirmultinomial.fit) fails.
61 Evidently, multiple "mlogit"s saved on vglm.dirmultinomial.fit@misc
62 do not suffice.
63
64
65
66
67 2011-12
68
69 VGAM version 0.8-4 said it needed R version 2-11.1 or later.
70 But really, R version 2-13.0 or later is needed.
71 This is because the generic nobs() was not defined properly.
72 Another fix is to install the (latest) prerelease version
73 at http://www.stat.auckland.ac.nz/~yee/VGAM/prerelease
74
75
76
77
78 2010-04-12
79 cqo() should be working now. It uses new C code.
80 Also, vgam() and vsmooth.spline() should not be noticeably different
81 from before. But cao() is still working... getting it going soon hopefully.
82
83
84
85
86 2009/07/13
87 cqo() fails... I think it is due to initial values being faulty.
88 Hope to look into it soon.
89
90
91
92
93 2009/06/18
94
95 For a given VGAM family function,
96 arguments such as parallel, exchangeable etc.
97 will not work if the RHS of the formula is an intercept only.
98 For example,
99 parallel = FALSE ~ 1
100 and
101 exchangeable = TRUE ~ 1
102 will fail.
103 Instead, try something like
104 parallel = TRUE ~ x2 + x3 + x4 -1
105 and
106 exchangeable = FAlSE ~ x2 + x3 + x4 + x5 -1
107 respectively.
108
109
110 2009/01/01
111
112 prediction with vgam( ~ offset(myoffsetmatrix) + ... )
113 fails inside a function because myoffsetmatrix cannot be found.
114
115
116
117 2008/08/12
118
119 Under Windows, the vgam() example involving the Hunua data seems
120 to fail. It is under investigation.
121
122
123
124 2008/08/04
125
126 VGAM interferes with other packages, e.g., predict() and summary().
127 This is due to S3 and S4 interference, and currently I haven't sussed out
128 the full details (e.g., NAMESPACES). For now it is best to attach VGAM
129 only when needed and detach it when other packages are to be used. This
130 can be done with
131
132 library(VGAM)
133
134 and
135
136 detach("package:VGAM")
137
138
139
140 2008/05/16
141
142 zipf() did not handle 0 < s < 1.
143 The prerelease version fixes this.
144
145
146
147 2008/03/12
148
149 A call such as
150
151 mydof = 4
152 Fit = vgam(y ~ s(x, df=mydof), fam=poissonff)
153
154 will result in failure when
155
156 plot(Fit)
157
158 Instead, one needs
159
160 Fit = vgam(y ~ s(x, df=4), fam=poissonff)
161
162
163
164 2008/02/16
165
166 The VGAM package interferes with other functions, for example, if VGAM
167 is loaded and lmobject is an "lm" object then
168
169 fitted(lmobject)
170 predict(lmobject)
171 resid(lmobject)
172 residuals(lmobject)
173
174 will fail.
175
176
177
178 2006/05/18
179
180 dirmul() is not working yet.
181
182
183
184
185 2005/11/16
186
187 cao() now works in Windows.
188 The argument xij does not work properly.
189
190
191
192 2005/8/31
193
194 The windows version of cao() seems to hang.
195 It does not hang in Linux.
196
197
198 2005/6/10
199
200 cao() works in Linux but seems to hang in Windows.
201 The latter (distributed in a .zip file format) is made from a R Cross
202 Build process which may be a reason for the bug. I'm slowly looking
203 into the bug.
204
205
206 2005/5/6
207
208 The VGAM package interferes with other code, including glm(). This may
209 be due to the smart prediction code, or be due to the NAMESPACE facility.
210 In order to use other functions outside the VGAM package you may need
211 to type "detach()".
212
213
214 2003/7/14
215
216 vgam(y ~ s(x, df=2), subset= x > 2)
217
218 will fail in R because the subset argument has the effect that the
219 "df" and "spar" attributes are removed from the data frame containing
220 the smoothing variables.
221
222 Current fix: create a separate data frame satisfying the subset=
223 condition, and then run vgam() on this smaller data frame.
224
225 Thanks for Eugene Zwane for finding this bug.
226
227
228