Feedback from my first impression of PyMC3 chapter
amcknight opened this issue · 2 comments
The language is very specific and clear.
There is a lot of assumed knowledge that made it difficult for me to follow some of it. Maybe be a bit more clear about what you assume the reader knows at the beginning of the chapter? Maybe this is covered in earlier chapters.
I would have liked many more links, specifically to terms you use. (for instance, no links to Theano, HMC, NUTS)
Some other little notes:
Confusing:
"After successfull installation, PyMC3 can be installed using pip:"
After successful installation of what?
"The final piece of the model, Y_obs, is an observed stochastic, and represents the data likelihood of the model"
I found this confusing but I'm not sure if it's just because of my unfamiliarity with some stats terminology. Is there a better way to say "data likelihood of the model"?
"Perhaps we are now trying to predict the values of Y that are greater than zero:"
Saying perhaps here makes it sound like we're guessing what's happening. A little clunky.
Assumes I know what GLM is
Should maybe put "(GLM)" after first use of general linear model rather than in its section near the end.
Conclusion ends abruptly, mid-sentence
Typos:
obivated
successfull
withing
"we have instantiate a step method"
"might not be possible to write a probability distributions"
"mathematical simplifications of the and caching"
I guess that's it. It seems very solid, though not very beginner friendly, but maybe that's fine given the context of the book.
Thanks for the feedback Andrew, I appreciate it a lot.
Would you mind expanding on which parts were difficult to follow? What things did you feel like were assumed?
Thanks @amcknight, that's very valuable and we'll improve those parts!