tumanian

Precision and recall are a great way of looking at overall accuracy, but ROC curves and ROC AUC are highly recommended for model selection criteria on imbalnced datasets. For SVMs you can do a two dimensional grid search for C and Gamma looking for a maximum ROC.

I also use per class weights - by default SVM(libSVM, libLinear) cost parameter is the same for both classes. Penalize the classifier for false negatives more then for false positives(order of magnitude more).

vi /etc/hosts

0.0.0.0 facebook.com 0.0.0.0 reddit.com 0.0.0.0 news.ycombinator.com

wq!

Hadoop API is a pain. Well, it is fine when doing the vanilla mapreduce, but once one steps away from that, things get ugly real fast with the zoo of Jobs, Tasks, Contexts, two mapred/mapreduce apis which do the same thing, plus hadoop.23 specific calls. I haven't touched Hadoop 1.0 yet, though I doubt they have cleaned all the mess. Hope they will streamline the API by 2.0

Devs are fine. We lost all our QE.

The C++ Programming Language and the ARM(Annotated Reference Manual) won me many interviews. After reading those I could claim that I know the language. Sadly, I could never find such concise reference for Python or Java (or Erlang, but there was no documentation anyway).

Check out this: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.122...

A great textbook as an intro is by Duda and Hart, Pattern Classification. http://www.amazon.com/Pattern-Classification-2nd-Richard-Dud.... Its pretty well written and gives a good overview of the main techniques. If you want a bit more theory, try Cherkassky and Muller, "Learning from Data".http://www.amazon.com/Learning-Data-Concepts-Theory-Methods/ Has a good overview section on statistical learning theory. And also, take WEKA and just play with it.Its nice to just check what works and what doesn't.

The article is correct in the insight part, however it misses(or doesn't state explicitly) the fact that insight comes from the definition of the problem, what you are trying to do. Every concept is a solution to a problem, and to get the insight, my approach is to go through a list What am I trying to accomplish (this is where visualizing comes) What are the other ways of doing it How does this method work, and why Why is this method better then others *Where will this method not work.

After this solving any problem in the problem set is a piece of cake. Reduce the problem to subproblems, check applicability of the concept to the subproblems, apply the method, enjoy the result.