bayesoptbook/bayesoptbook.github.io

Add datasets for replication

Opened this issue · 4 comments

Description of issue
Can you add some of the toy data sets you used to generate the visuals. .

Proposed resolution
Add data sets per chapter

Additional context
It would help readers solidify their understanding if they could try and replicate the work in the book

Additional context
Add any other context or screenshots about the feature request here.

Did you have a particular dataset (or datasets) you were interested in ? Happy to post what I can here.

To be honest no, i didn't have a specific data set in mind. I like to read a book like yours and replicate what I can to be sure I understood the technique. Was struck that the lack of data sets made it harder to engage with your book.

dothit commented

Maybe if I may add to the proposal with concrete examples of data sets I would require and like to use to reproduce your results:

  • Observations and their variances, as well as model structures to model noisy processes in Figs. 2.3 and 2.4.
  • Dataset shown in Fig. 4.2 and the model structure as well as the range of sigma and ell that leads to Fig. 4.3.
  • Values of objective function in Fig. 7.2 and the initial observations and model structure that are used to produce the various acquisition functions in Figs. 7.3 (EI), 7.5 (KG), 7.9 (PI), 7.14 (MI), 7.17 (UCB), and 7.23 (MEI).
  • As well as the noisy variants in chapter 8, but I would like to follow up on that after reading the chapter again.

This would include the datasets themselves, as well as the results, if possible. Now that I write it out, I'm kind of begging for a lot. But I guess there is no harm in asking ;-)

I should be able to extract at least some of this! It might take me a little while but I will try to find the time.