Jesse Cambon 12 April, 2020
References: * https://tidymodels.github.io/parsnip/articles/parsnip_Intro.html * https://www.tidyverse.org/blog/2018/11/parsnip-0-0-1/ * https://m-clark.github.io/workshops/bayesian/04_R.html
Requires packages: ranger, randomForest, rstanarm
library(parsnip)
library(tidymodels)## ── Attaching packages ────────────────────────────────────────────────────────────────────── tidymodels 0.1.0 ──
## ✓ broom 0.5.5 ✓ recipes 0.1.9
## ✓ dials 0.0.6 ✓ rsample 0.0.6
## ✓ dplyr 0.8.5 ✓ tibble 2.1.3
## ✓ ggplot2 3.3.0 ✓ tune 0.1.0
## ✓ infer 0.5.1 ✓ workflows 0.1.1
## ✓ purrr 0.3.3 ✓ yardstick 0.0.6
## ── Conflicts ───────────────────────────────────────────────────────────────────────── tidymodels_conflicts() ──
## x purrr::discard() masks scales::discard()
## x dplyr::filter() masks stats::filter()
## x recipes::fixed() masks stringr::fixed()
## x dplyr::lag() masks stats::lag()
## x ggplot2::margin() masks dials::margin()
## x recipes::step() masks stats::step()
## x recipes::yj_trans() masks scales::yj_trans()
set.seed(4831) # set seed for reproducibility
# Split data
split <- initial_split(mtcars, props = 9/10)
car_train <- training(split)
car_test <- testing(split)
# Defines type of model we want
car_model <- linear_reg()library(rstanarm)## Loading required package: Rcpp
##
## Attaching package: 'Rcpp'
## The following object is masked from 'package:rsample':
##
## populate
## rstanarm (Version 2.19.3, packaged: 2020-02-11 05:16:41 UTC)
## - Do not expect the default priors to remain the same in future rstanarm versions.
## Thus, R scripts should specify priors explicitly, even if they are just the defaults.
## - For execution on a local, multicore CPU with excess RAM we recommend calling
## options(mc.cores = parallel::detectCores())
## - bayesplot theme set to bayesplot::theme_default()
## * Does _not_ affect other ggplot2 plots
## * See ?bayesplot_theme_set for details on theme setting
wide_prior <- normal(0, 10)
stan_car_model <-
car_model %>%
set_engine("stan", iter = 5000, prior = wide_prior, seed = 2347)
stan_car_model## Linear Regression Model Specification (regression)
##
## Engine-Specific Arguments:
## iter = 5000
## prior = wide_prior
## seed = 2347
##
## Computational engine: stan
# don't print anything:
ctrl <- fit_control(verbosity = 0)
stan_fit <-
stan_car_model %>%
fit(mpg ~ ., data = car_train, control = ctrl)
stan_fit## parsnip model object
##
## Fit time: 6s
## stan_glm
## family: gaussian [identity]
## formula: mpg ~ .
## observations: 24
## predictors: 11
## ------
## Median MAD_SD
## (Intercept) -10.9 32.4
## cyl 0.8 1.9
## disp 0.0 0.0
## hp 0.0 0.0
## drat 2.4 2.3
## wt -3.3 2.3
## qsec 1.0 0.9
## vs 1.5 2.9
## am 3.4 2.8
## gear 2.6 2.8
## carb -0.9 1.3
##
## Auxiliary parameter(s):
## Median MAD_SD
## sigma 3.0 0.6
##
## ------
## * For help interpreting the printed output see ?print.stanreg
## * For info on the priors used see ?prior_summary.stanreg
predict(stan_fit, car_test,type='conf_int')## # A tibble: 8 x 2
## .pred_lower .pred_upper
## <dbl> <dbl>
## 1 17.2 26.1
## 2 11.1 18.9
## 3 11.2 19.3
## 4 6.95 18.1
## 5 13.2 29.4
## 6 13.5 21.7
## 7 26.0 31.3
## 8 12.5 36.4
library(tidybayes)##
## Attaching package: 'tidybayes'
## The following object is masked from 'package:tune':
##
## parameters
## The following object is masked from 'package:dials':
##
## parameters
library(bayesplot)## This is bayesplot version 1.7.1
## - Online documentation and vignettes at mc-stan.org/bayesplot
## - bayesplot theme set to bayesplot::theme_default()
## * Does _not_ affect other ggplot2 plots
## * See ?bayesplot_theme_set for details on theme setting
# tidybayes to extract info
stan_fit$fit %>% get_variables()## [1] "(Intercept)" "cyl" "disp" "hp"
## [5] "drat" "wt" "qsec" "vs"
## [9] "am" "gear" "carb" "sigma"
## [13] "accept_stat__" "stepsize__" "treedepth__" "n_leapfrog__"
## [17] "divergent__" "energy__"
## Use bayesplot to plot
posterior <- as.matrix(stan_fit$fit)
plot_title <- ggtitle("Posterior distributions",
"with medians and 80% intervals")
mcmc_areas(posterior,
pars = c("cyl", "drat", "am", "wt"),
prob = .8) + plot_title## Warning: `expand_scale()` is deprecated; use `expansion()` instead.
#stan_fit %>% spread_draws()pp_check(stan_fit$fit)
library(shinystan)## Loading required package: shiny
##
## This is shinystan version 2.5.0
launch_shinystan(stan_fit$fit)##
## Hang on... preparing graphical posterior predictive checks for rstanarm model.
## See help('shinystan', 'rstanarm') for how to disable this feature.
##
## Launching ShinyStan interface... for large models this may take some time.
##
## Listening on http://127.0.0.1:6429
rf_with_seed <-
rand_forest(trees = 2000, mtry = varying(), mode = "regression") %>%
set_engine("ranger", seed = 63233)
# Fig with ranger
ranger_model <- rf_with_seed %>%
set_args(mtry = 4) %>%
set_engine("ranger",keep.inbag=TRUE) %>%
fit(mpg ~ ., data = mtcars)
# First with random forest package
rf_model <- rf_with_seed %>%
set_args(mtry = 4) %>%
set_engine("randomForest") %>%
fit(mpg ~ ., data = mtcars)ranger_predictions <- predict(ranger_model, mtcars, type = "conf_int")