Comparing models with WAIC
Now that you have successfully built the first, basic model, you take another look at the data at your disposal. You notice a variable called wind_speed
. This could be a great predictor of the numbers of bikes rented! Cycling against the wind is not that much fun, is it?
You fit another model with this additional predictor:
formula = "num_bikes ~ temp + work_day + wind_speed"
with pm.Model() as model_2:
pm.GLM.from_formula(formula, data=bikes)
trace_2 = pm.sample(draws=1000, tune=500)
Is your new model_2
better than model_1
, the one without wind speed? Compare the two models using Widely Applicable Information Criterion, or WAIC, to find out!
Both trace_1
and trace_2
are available in your workspace, and pycm3
has been imported as pm
.
This exercise is part of the course
Bayesian Data Analysis in Python
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Gather trace_1 and trace_2 into a dictionary
traces_dict = ____