Greetings all,
I have a rather strange (may not be strange and it is my inexperiencedness) issue. After sampling via rstamarm, I use parameters package’s model_parameters to observe results. However, when I re-run the model for the second without any changes, I got;
Warning messages:
1: There were 1 chains where the estimated Bayesian Fraction of Missing Information was low. See
https://mc-stan.org/misc/warnings.html#bfmi-low
2: Examine the pairs() plot to diagnose sampling pro
Is it a problem that I should be concerned?
m1 <- stan_glm(n~ -1 + Native_Language, family = neg_binomial_2(link = 'log'), init_r = 0.5, QR = TRUE, iter = 10000, prior_intercept = normal(2, 0.5), prior = normal(0, 2.5, autoscale = TRUE), data = alternation_freq)
> m1
stan_glm
family: neg_binomial_2 [log]
formula: n ~ -1 + Native_Language
observations: 27
predictors: 27
------
Median MAD_SD
Native_LanguageBulgarian 5.1 0.9
Native_LanguageChinese 4.7 0.9
Native_LanguageChinese-Cantonese 6.1 0.9
Native_LanguageCzech 5.1 0.9
Native_LanguageDutch 5.3 0.9
Native_LanguageFinnish 4.6 0.9
Native_LanguageFrench 4.9 0.9
Native_LanguageGerman 5.4 0.9
Native_LanguageGreek 5.4 0.9
Native_LanguageHungarian 4.9 0.9
Native_LanguageItalian 5.1 0.9
Native_LanguageJapanese 5.6 0.9
Native_LanguageKorean 5.1 0.9
Native_LanguageLithuanian 4.6 0.9
Native_LanguageMacedonian 5.1 0.9
Native_LanguageNorwegian 5.3 0.9
Native_LanguagePersian 5.1 0.9
Native_LanguagePolish 5.0 0.9
Native_LanguagePortuguese 5.0 0.9
Native_LanguagePunjabi 5.2 0.9
Native_LanguageRussian 5.5 0.9
Native_LanguageSerbian 5.3 0.9
Native_LanguageSpanish 5.0 0.9
Native_LanguageSwedish 5.4 0.9
Native_LanguageTswana 5.4 0.9
Native_LanguageTurkish 5.6 0.9
Native_LanguageUrdu 4.8 0.9
Auxiliary parameter(s):
Median MAD_SD
reciprocal_dispersion 1.4 0.9
------
* For help interpreting the printed output see ?print.stanreg
* For info on the priors used see ?prior_summary.stanreg
So far so good. However;
x1_1 <- model_parameters(m1, centrality = "median", dispersion = FALSE, ci = .89, ci_method = "hdi", test = c("pd","ROPE", 'BF'),exponentiate = FALSE, diagnostic = "Rhat", priors = FALSE)
Sampling priors, please wait...
> x1_1
Parameter | Median | 89% CI | pd | % in ROPE | Rhat | BF
--------------------------------------------------------------------------------------------
Native_LanguageBulgarian | 5.06 | [3.59, 6.83] | 100% | 0% | 1.000 | > 1000
Native_LanguageChinese | 4.72 | [3.22, 6.48] | 100% | 0% | 1.000 | > 1000
Native_LanguageChinese-Cantonese | 6.14 | [4.61, 7.90] | 100% | 0% | 1.000 | > 1000
Native_LanguageCzech | 5.08 | [3.58, 6.93] | 100% | 0% | 1.000 | > 1000
Native_LanguageDutch | 5.29 | [3.79, 7.13] | 100% | 0% | 1.000 | > 1000
Native_LanguageFinnish | 4.61 | [3.16, 6.48] | 100% | 0% | 1.000 | > 1000
Native_LanguageFrench | 4.94 | [3.42, 6.74] | 100% | 0% | 1.000 | > 1000
Native_LanguageGerman | 5.39 | [3.84, 7.19] | 100% | 0% | 1.001 | > 1000
Native_LanguageGreek | 5.37 | [3.90, 7.22] | 100% | 0% | 1.001 | > 1000
Native_LanguageHungarian | 4.87 | [3.32, 6.58] | 100% | 0% | 1.000 | > 1000
Native_LanguageItalian | 5.07 | [3.60, 6.82] | 100% | 0% | 1.000 | > 1000
Native_LanguageJapanese | 5.58 | [4.11, 7.41] | 100% | 0% | 1.000 | > 1000
Native_LanguageKorean | 5.12 | [3.61, 6.93] | 100% | 0% | 1.000 | > 1000
Native_LanguageLithuanian | 4.60 | [3.11, 6.47] | 100% | 0% | 1.000 | > 1000
Native_LanguageMacedonian | 5.09 | [3.64, 6.97] | 100% | 0% | 1.000 | > 1000
Native_LanguageNorwegian | 5.26 | [3.75, 6.98] | 100% | 0% | 1.000 | > 1000
Native_LanguagePersian | 5.08 | [3.63, 6.92] | 100% | 0% | 1.000 | > 1000
Native_LanguagePolish | 5.02 | [3.55, 6.81] | 100% | 0% | 1.000 | > 1000
Native_LanguagePortuguese | 5.01 | [3.52, 6.79] | 100% | 0% | 1.000 | > 1000
Native_LanguagePunjabi | 5.15 | [3.71, 7.01] | 100% | 0% | 1.000 | > 1000
Native_LanguageRussian | 5.47 | [4.00, 7.24] | 100% | 0% | 1.000 | > 1000
Native_LanguageSerbian | 5.31 | [3.82, 7.12] | 100% | 0% | 1.000 | > 1000
Native_LanguageSpanish | 5.02 | [3.57, 6.82] | 100% | 0% | 1.000 | > 1000
Native_LanguageSwedish | 5.41 | [3.97, 7.25] | 100% | 0% | 1.000 | > 1000
Native_LanguageTswana | 5.40 | [4.01, 7.25] | 100% | 0% | 1.000 | > 1000
Native_LanguageTurkish | 5.56 | [4.09, 7.42] | 100% | 0% | 1.000 | > 1000
Native_LanguageUrdu | 4.77 | [3.30, 6.61] | 100% | 0% | 1.000 | > 1000
reciprocal_dispersion | 1.42 | [0.27, 3.12] | 100% | 0% | 1.001 | > 1000
Uncertainty intervals (highest-density) and p values (two-tailed) computed using a MCMC distribution approximation.
> m1 <- stan_glm(n~ -1 + Native_Language, family = neg_binomial_2(link = 'log'), init_r = 0.5, QR = TRUE, iter = 10000, prior_intercept = normal(2, 0.5), prior = normal(0, 2.5, autoscale = TRUE), data = alternation_freq)
Warning messages:
1: There were 1 chains where the estimated Bayesian Fraction of Missing Information was low. See
https://mc-stan.org/misc/warnings.html#bfmi-low
2: Examine the pairs() plot to diagnose sampling problems
**sessionInfo()**
R version 4.1.1 (2021-08-10)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 22000)
Matrix products: default
locale:
[1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252 LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C LC_TIME=English_United States.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] bayestestR_0.11.5 parameters_0.17.0 rstanarm_2.21.1 Rcpp_1.0.8.3 forcats_0.5.1 stringr_1.4.0 dplyr_1.0.8
[8] purrr_0.3.4 readr_2.1.2 tidyr_1.2.0 tibble_3.1.6 ggplot2_3.3.5 tidyverse_1.3.1 readxl_1.3.1