Simulating posterior draws
You have just decided to use a Beta(5, 2) prior for the efficacy rate. You are also using the binomial distribution to model the data (curing a sick patient is a "success", remember?). Since the beta distribution is a conjugate prior for the binomial likelihood, you can simply simulate the posterior!
You know that if the prior is \(Beta(a, b)\), then the posterior is \(Beta(x, y)\), with:
\(x = NumberOfSuccesses + a\),
\(y = NumberOfObservations - NumberOfSuccesses + b\).
Can you simulate the posterior distribution? Recall that altogether you have data on 22 patients, 19 of whom have been cured. numpy
and seaborn
have been imported for you as np
and sns
, respectively.
Diese Übung ist Teil des Kurses
Bayesian Data Analysis in Python
Anleitung zur Übung
- Assign the numbers of patients treated and cured to
num_patients_treated
andnum_patients_cured
, respectively. - Use the appropriate
numpy
function to sample from the posterior distribution and assign the result toposterior_draws
. - Plot the posterior distribution using the appropriate
seaborn
function.
Interaktive Übung
Versuche dich an dieser Übung, indem du diesen Beispielcode vervollständigst.
# Define the number of patients treated and cured
num_patients_treated = ____
num_patients_cured = ____
# Simulate 10000 draws from the posterior distribuition
posterior_draws = ____(____ + ____, ____ - ____ + ____, 10000)
# Plot the posterior distribution
____(____, shade=True)
plt.show()