my stats question is as follows:

The research group I work for have developed a theoretical growth model for a particular species of fish. The idea is that if you provide some initial starting values for the model you then generate an expected growth curve along with 95% confidence bands. To extend the model we would like to be able to update the model and recalculate the curve when real data becomes available. For example, imagine that at age 6 the model predicts an average weight of 34g, but in a random sample of fish we find that the mean age is, say, 24g, we would like to be able to use this new data to 'tweak' our previously estimated curve. At the moment I am not sure quite how to address this problem so any suggestions would be much appreciated.


  • 1
    $\begingroup$ just a quick idea: it seems for this question to belong to the realm of Bayesian approach (so may be there will be such a reply from an expert in that field). From the point of view of a regular $OLS$ user, why not to re-run the same estimation with a new dataset (it is hardly believable that calculations take too long time, though Bayesian tricks could be quicker here)? $\endgroup$ – Dmitrij Celov Mar 21 '11 at 16:07

You could rewrite your model in a Bayesian software (OpenBUGS, PyMC). When any new information is available add them to the model and re-estimate the posterior.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.