nlme issue since 3.40

Wed, May 10, 2017 8:29 PM

Working with Win10 64 bit. Same point estimates as Tom.

R version 3.4.0 (2017-04-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 14393)

Matrix products: default


Best,
Jinsong

On 2017/5/11 4:54, Philippi, Tom wrote:

Jochen--

Using R 3.4 and nlme 3.1-131 on MS win7 and the data you included in the
other message thread, I cannot duplicate your error.  It also runs fine
under MRAN 3.3.2 on MS win7 which uses the Intel MKL.  I haven't yet
updated R to 3.4 on my linux ubuntu box to test there, but my best guess
now is that something failed in your update to 3.4.

library(nlme)
library(broom)

data2 <- structure(list(Place = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L,

+ 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L,
+ 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L,
+ 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label =
+ c("Cambridge Bay",
+ "Iqaluit", "Rankin Inlet"), class = "factor"), year_c = c(-2,
+ -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, -2,
+ -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, -2,
+ -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15), SR_Total_ctd
+ = c(NA,
+ NA, 5.48961420000001, 7.24637679999999, 3.76569040000001, 6.9014085,
+ 9.3088858, 5.1771117, 7.51366120000002, 7.18157180000001, 3.7275064,
+ 5.7397959, 6.87500000000001, 6.25, 5.84577109999999, 6.2421973,
+ 4.70446319999999, NA, NA, NA, 11.9192688, 9.6291013, 8.54640980000001,
+ 8.7893297, 7.42343539999999, 5, 5.3674121, 6.1322261, 6.4182843,
+ 5.75692960000001, 6.3861534, 6.6706021, 7.76642339999999,
6.91983119999999,
+ 5.3634631, NA, NA, NA, 0.0870321999999959, -1.19047620000001,
+ -1.8723404, -0.4244482, 1.18443319999999, 2.19594590000001,
+ 1.57545610000001,
+ 3.080766, 1.8744906, 5.51053480000002, 6.31412790000002, 7.4451411,
+ 8.1226054, 9.52743899999999, 9.4339623, NA)), .Names = c("Place",
+ "year_c", "SR_Total_ctd"), row.names = c(NA, -54L), class = "data.frame")


uncmod <- lme(fixed = year_c~1, random = ~1|Place, data = data2,

method="ML")

tidy(uncmod)

  group         level        term estimate
1 Place Cambridge Bay (Intercept)      6.5
2 Place       Iqaluit (Intercept)      6.5
3 Place  Rankin Inlet (Intercept)      6.5

summary(uncmod)

Linear mixed-effects model fit by maximum likelihood
 Data: data2
       AIC      BIC    logLik
  337.0536 343.0206 -165.5268

Random effects:
 Formula: ~1 | Place
         (Intercept) Residual
StdDev: 0.0001350644 5.188127

Fixed effects: year_c ~ 1
            Value Std.Error DF  t-value p-value
(Intercept)   6.5 0.7126441 51 9.120962       0

Standardized Within-Group Residuals:
          Min            Q1           Med            Q3           Max
-1.638356e+00 -8.673650e-01 -1.193490e-15  8.673650e-01  1.638356e+00

Number of Observations: 54
Number of Groups: 3

VarCorr(uncmod)

Place = pdLogChol(1)
            Variance     StdDev
(Intercept) 1.824240e-08 0.0001350644
Residual    2.691667e+01 5.1881274721

sessionInfo()

R version 3.4.0 (2017-04-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252
[2] LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

other attached packages:
[1] broom_0.4.2  nlme_3.1-131

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.10     lattice_0.20-35  tidyr_0.6.1      psych_1.7.3.21
 [5] dplyr_0.5.0      assertthat_0.2.0 grid_3.4.0       R6_2.2.0
 [9] plyr_1.8.4       DBI_0.6-1        magrittr_1.5     stringi_1.1.5
[13] lazyeval_0.2.0   reshape2_1.4.2   tools_3.4.0      stringr_1.2.0
[17] foreign_0.8-67   parallel_3.4.0   compiler_3.4.0   mnormt_1.5-5
[21] tibble_1.3.0


Tom 2


On Wed, May 10, 2017 at 1:05 PM, Jochen Wirsing <jw1085 at wildcats.unh.edu>
wrote:

Dear All,


thanks again for your nice feedback. The Session Info etc was mentioned
in the previous email, but one of the folks here on the list asked me to
use dput() so I could paste my data in textform here.

So here you go:

structure(list(Place = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L,
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L,
3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label =
c("Cambridge Bay",
"Iqaluit", "Rankin Inlet"), class = "factor"), year_c = c(-2,
-1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, -2,
-1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, -2,
-1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15), SR_Total_ctd
= c(NA,
NA, 5.48961420000001, 7.24637679999999, 3.76569040000001, 6.9014085,
9.3088858, 5.1771117, 7.51366120000002, 7.18157180000001, 3.7275064,
5.7397959, 6.87500000000001, 6.25, 5.84577109999999, 6.2421973,
4.70446319999999, NA, NA, NA, 11.9192688, 9.6291013, 8.54640980000001,
8.7893297, 7.42343539999999, 5, 5.3674121, 6.1322261, 6.4182843,
5.75692960000001, 6.3861534, 6.6706021, 7.76642339999999, 6.91983119999999,
5.3634631, NA, NA, NA, 0.0870321999999959, -1.19047620000001,
-1.8723404, -0.4244482, 1.18443319999999, 2.19594590000001,
1.57545610000001,
3.080766, 1.8744906, 5.51053480000002, 6.31412790000002, 7.4451411,
8.1226054, 9.52743899999999, 9.4339623, NA)), .Names = c("Place",
"year_c", "SR_Total_ctd"), row.names = c(NA, -54L), class = "data.frame")


I was also asked, whether using lme4 would be an option, but I don't
think that this will help fix the issue that I believe occurred through
updating to R 3.40. Also, nlme provides some information in a nicer way
(covariances etc, if I remember correctly - all this is still very new
to me....), so I'd really rather see nlme fixed.


Thanks again!

nlme issue since 3.40

Thread (5 messages)