I wrote a blog post about predicting my dog’s weight using Bayesian modeling with pymc and @twiecki suggested I share it here, too.
This was a great read. Do these models work for the lifetime of a dog or do they end up predicting
I wonder why do you think that the dog would not loose weight while becoming old?
Good points! These models will asymptotically approach Y_M, so they shouldn’t grow to the size of a whale. I think if you have data that includes the change in weight starting to slow down, you should arrive at a reasonable maximum weight. If your data only includes growth, who knows how wrong a least squares method might be. But the Bayesian method should be grounded by the Y_M prior.
A more useful question might be “How tall/long/wide will my dog grow to be?” For me, weight is a proxy for dog size, because it is much easier to measure. There is some final size a dog may reach, but the weight will change some as food/activity/age changes. Depending on the dog and owner, the dog might put on extra weight as they grow old!
are you talking about the BMI for dogs?