8.57 Evaluate the model

… and we take a look at the testing performance (notice an improvement over the decision tree).

augment(bagging_fit, new_data = carseats_train) %>%
  rmse(truth = Sales, estimate = .pred)

## # A tibble: 1 × 3
##   .metric .estimator .estimate
##   <chr>   <chr>          <dbl>
## 1 rmse    standard       0.671

augment(bagging_fit, new_data = carseats_test) %>%
  rmse(truth = Sales, estimate = .pred)

## # A tibble: 1 × 3
##   .metric .estimator .estimate
##   <chr>   <chr>          <dbl>
## 1 rmse    standard        1.35

Training RMSE: 0.671 Testing RMSE: 1.35 (overfit)

We can also create a quick scatterplot between the true and predicted value to see if we can make any diagnostics.

augment(bagging_fit, new_data = carseats_test) %>% 
     ggplot(aes(Sales, .pred)) + 
     geom_abline() + 
     geom_point(alpha = 0.5)