The exam exercises for this chapter were much trickier than the homework exercises. With the right approach, they could be solved quickly; too bad I never found that right approach…
It has been a long time since I wrote about this book; I had worked the solutions more than a month ago, but then life happened, and I could not find the time (or, perhaps, more accurately the courage) to typeset my notes…
Anyway, I do have time now and am eager to go on with Chapter 3; but first let’s finish Chapter 2. Today the homework exercises, and very soon the exams and bonus (at least the ones I could do) exercises.
I am currently reading Machine Learning in Action, as I need something light between sessions with Concrete Mathematics. This book introduces a number of important machine learning algorithms, each time with a complete implementation and one or more test data sets; it also explains the underlying mathematics, and provides information about additional reference material (mostly heavier and more expensive books).
However, in Chapter 4 about Naïve Bayes classifiers, I didn’t see how the implementation derived by the maths. Eventually, I confirm that it could not, and try to correct it.
This has lasted a little bit longer than seven weeks (the release schedule of the beta versions did not help; my day job did not help either), but finally I finished the book.
Wow, almost two months since I wrote Day 2, and more than one since the last post in this series… Time to bring it to an end.
This second batch of exercises builds on the previous one. Once again, there are no complex manipulations, and very often the solution just follows from the definitions.
This first batch of exercises is meant to develop familiarity with the various concepts and notations introduced in this chapter. There is no complex manipulation, but the trick is to be aware of the often unmentioned assumptions about the precise meaning of the expressions.
After a long but busy silence, I have now a few notes on the second chapter, Sums. As with Chapter 1, these are nothing revolutionary; just some clarifications of the points that were not obvious to me, as well as other, random observations.
In Psychic Modeling, I described a reasonably understandable implementation of a ticket generator for the Psychic Modeling Problem. While this version is not overly slow, it is not amazingly fast either.
As I’m refreshing my C skills, I thought it would be interesting to try and implement a version as fast as possible.