Visualize the xgboost bike rental model
You've now seen three different ways to model the bike rental data. For this example, you've seen that the gradient boosting model had the smallest RMSE. To finish up the course, let's compare the gradient boosting model's predictions to the other two models as a function of time.
On completing this exercise, you will have completed the course. Congratulations! Now you have the tools to apply a variety of approaches to your regression tasks.
The data frame bikesAugust
with predictions,has been pre-loaded. The plots quasipoisson_plot
and randomforest_plot
are also available.
This exercise is part of the course
Supervised Learning in R: Regression
Exercise instructions
- Print
quasipoisson_plot
to review the quasipoisson model's behavior. - Print
randomforest_plot
to review the random forest model's behavior. - Fill in the blanks to plot the gradient boosting predictions and actual counts by hour for the first 14 days of August.
pivot_longer()
thecnt
andgbm
column names into a column calledvalue
, with a key calledvaluetype
.- Plot
value
as a function ofinstant
(day).
How does the gradient boosting model compare to the previous models?
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Print quasipoisson_plot
___
# Print randomforest_plot
___
# Plot predictions and actual bike rentals as a function of time (days)
bikesAugust %>%
mutate(instant = (instant - min(instant))/24) %>% # set start to 0, convert unit to days
filter(instant < 14) %>% # first two weeks
pivot_longer(c(___, ___), names_to = ___, values_to = ___) %>%
ggplot(aes(x = ___, y = ___, color = valuetype, linetype = valuetype)) +
geom_point() +
geom_line() +
scale_x_continuous("Day", breaks = 0:14, labels = 0:14) +
scale_color_brewer(palette = "Dark2") +
ggtitle("Predicted August bike rentals, Gradient Boosting model")