Closed wpetry closed 4 years ago
augment.brmsfit() returns predictions without the associated data/newdata columns.
augment.brmsfit()
Reproducible example:
library(brms) library(broom.mixed) library(tibble) library(tidyr) # fit model (from brm() examples) bprior1 <- prior(student_t(5,0,10), class = b) + prior(cauchy(0,2), class = sd) fit1 <- brm(count ~ zAge + zBase * Trt + (1|patient), data = epilepsy, family = poisson(), prior = bprior1) augment(fit1) # only returns columns .fitted, .se.fit, .resid augment(fit1, newdata = tidyr::crossing(zAge = 0, zBase = 0, Trt = 0, patient = 1:10)) # only returns columns .fitted, .se.fit
Expected output when not supplying new data:
bind_cols(as_tibble(fit1$data), augment(fit1))
# A tibble: 236 x 8 count zAge zBase Trt patient .fitted <dbl> <dbl> <dbl> <fct> <fct> <dbl> 1 5 0.425 -0.757 0 1 3.52 2 3 0.265 -0.757 0 2 3.55 3 2 -0.533 -0.944 0 3 2.76 4 4 1.22 -0.870 0 4 3.30 5 7 -1.01 1.30 0 5 13.7 6 5 0.106 -0.158 0 6 5.48 7 6 0.425 -0.720 0 7 3.15 8 40 2.18 0.778 0 8 22.7 9 5 1.38 -0.308 0 9 5.50 10 14 -0.0541 -0.795 0 10 7.67 # … with 226 more rows, and 2 more variables: # .se.fit <dbl>, .resid <dbl>
augment.brmsfit()
returns predictions without the associated data/newdata columns.Reproducible example:
Expected output when not supplying new data: