Exercise

Clicks per day instead of clicks per ad

When you put up a banner on your friend's site you got 19 clicks in a day, how many daily clicks should you expect this banner to generate on average? Now, modify your model, one piece at a time, to calculate this.

Instructions 1/4

undefined XP
    1
    2
    3
    4

To the right is the old code for the binomial Bayesian model. To accommodate the new banner data you're now going to change the model so that it uses a Poisson distribution instead.

  • Start by replacing the prior distribution over proportion_clicks by a prior over mean_clicks. Make the prior a uniform (runif) distribution from 0 to 80 clicks per day.