Compatibility between phylolm and visreg
Closed this issue · 2 comments
Hi! Congratulations for the package! And thank you for maintaining it!
I've tried to visualize phylolm objects with the visreg package: the visualization lacks the confidence band around the slope. This lack of features could arise from the fact that model.frame() appears to return a string "lambda", rather than a data frame, when applied to phylolm objects. Could this be fixed? I think that compatibility with visreg would be a useful feature in future releases of phylolm.
See a reproducible example below. Cheers!
library(ape)
library(ade4)
library(phylolm)
library(visreg)
# use dataset from ade4 package: life-history traits for 18 lizards
data(lizards)
# phylogenetic tree
l.tree <- read.tree(text = lizards$hprA)
# traits
l.traits <- lizards$traits
# Fit pgls
mod <- phylolm::phylolm(matur.L ~ age.mat, data = l.traits, phy = l.tree, model = "lambda")
#> Warning in phylolm::phylolm(matur.L ~ age.mat, data = l.traits, phy = l.tree, : the estimation of lambda matches the upper/lower bound for this parameter.
#> You may change the bounds using options "upper.bound" and "lower.bound".
summary(mod)
#>
#> Call:
#> phylolm::phylolm(formula = matur.L ~ age.mat, data = l.traits,
#> phy = l.tree, model = "lambda")
#>
#> AIC logLik
#> 153.23 -72.62
#>
#> Raw residuals:
#> Min 1Q Median 3Q Max
#> -18.233 -13.593 1.759 6.276 31.485
#>
#> Mean tip height: 33
#> Parameter estimate(s) using ML:
#> lambda : 1e-07
#> sigma2: 5.663128
#>
#> Coefficients:
#> Estimate StdErr t.value p.value
#> (Intercept) 4.17891 11.37494 0.3674 0.7181507
#> age.mat 5.07030 0.99637 5.0888 0.0001095 ***
#> ---
#> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#>
#> R-squared: 0.6181 Adjusted R-squared: 0.5942
#>
#> Note: p-values and R-squared are conditional on lambda=1e-07.
# model.frame returns a string: "lambda"
model.frame(mod)
#> [1] "lambda"
# if I understand correctly, it should return something like this
model.frame(lm(matur.L ~ age.mat, data = l.traits))
#> matur.L age.mat
#> Sa 58 13
#> Sh 42 5
#> Tl 132 19
#> Mc 56 11
#> My 60 10
#> Pb 44 8
#> Ph 39 8
#> Pg 49 8
#> Pa 53 7
#> Pm 49 12
#> Tt 46 7
#> Ts 56 13
#> Ae 61 10
#> Zv 44 11
#> Zo 43 11
#> La 62 15
#> Ls 91 16
#> Lv 84 12
# formula appears to work fine
formula(mod)
#> matur.L ~ age.mat
# a visualization without confidence band
visreg(mod)
sessionInfo()
#> R version 4.2.0 (2022-04-22 ucrt)
#> Platform: x86_64-w64-mingw32/x64 (64-bit)
#> Running under: Windows 10 x64 (build 22000)
#>
#> Matrix products: default
#>
#> locale:
#> [1] LC_COLLATE=Spanish_Spain.utf8 LC_CTYPE=Spanish_Spain.utf8
#> [3] LC_MONETARY=Spanish_Spain.utf8 LC_NUMERIC=C
#> [5] LC_TIME=Spanish_Spain.utf8
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> other attached packages:
#> [1] visreg_2.7.0 phylolm_2.6.2 ade4_1.7-19 ape_5.6-2
#>
#> loaded via a namespace (and not attached):
#> [1] Rcpp_1.0.10 parallelly_1.31.1 rstudioapi_0.13 knitr_1.39
#> [5] magrittr_2.0.3 MASS_7.3-57 lattice_0.20-45 rlang_1.1.1
#> [9] fastmap_1.1.0 stringr_1.5.0 highr_0.9 globals_0.15.0
#> [13] tools_4.2.0 parallel_4.2.0 grid_4.2.0 nlme_3.1-157
#> [17] xfun_0.40 cli_3.6.1 withr_2.5.0 htmltools_0.5.6
#> [21] yaml_2.3.5 digest_0.6.29 lifecycle_1.0.3 codetools_0.2-18
#> [25] vctrs_0.6.3 fs_1.5.2 glue_1.6.2 evaluate_0.15
#> [29] rmarkdown_2.14 reprex_2.0.2 future.apply_1.9.0 stringi_1.7.6
#> [33] compiler_4.2.0 future_1.25.0 listenv_0.8.0Created on 2023-09-28 with reprex v2.0.2
Hello @MarcRieraDominguez ,
The new version of phylolm should be compatible with visreg. If you note any problems, please let me know.
Best,
Lam
Hello @lamho86
That's great! I've just tested the reprex with phylolm ver. 2.6.5, and visreg now returns a confidence band! Thank you very much!
I'll be closing this.