Passion is like genius; a miracle.

For R, use homebrew. Don’t miss –with-openblas. In my own experience, using –with-openblas improved performance x20 on a benchmark. To install scikit-learn, use homebrew and pip. Again, –with-openblas improved performance x2 for my program.

http://r4pda.co.kr/ 에서 확인하실 수 있습니다. 언제부터인지 정확히 기억은 나지 않지만, 이 책을 쓰는 일이 벌써 대략 일년이나 된 작업이 되었습니다. 그간 총 4번에 걸쳐 파일을 업로드했고 매번 3개월 정도의 시간이 걸렸기 때문입니다. 그간 몇분들로부터 내용에 대한 피드백을 받기도 했고, 공유한 내용에 자체에 대해 감사 메일을 주신 분도 계셨습니다. 저로서도 그분들께, 그리고 별다른 말씀없으셨어도 다운받아 가신…

This is great summary on the features selection: http://files.meetup.com/1225993/Deng_12_12_11_BARUG.pptx. It’s written by Houtao Deng who wrote Feature Selection via Regularized Trees(RRF). Here’s brief summary of the presentation: * Feature Selection Method – Univariate Filter: Information Gain, Chi-Squared – Multivariate Filter: CFS, FCBF – Wrapper Methods: LASSO, SVM-RFE, RF-RFE, RRF * Feature Selection Methods for Classifiers…

Problem to discuss is how to convert the following list of data frames into one data frame. One commonly known approach is to convert the list to vector using unlist(), and then convert to matrix, and then finally make it as data frame. But this has drawback as unlist() will perform type-coercion as vector can…

Here’s an example to load json objects into R’s data frame. You need some formatting of the json file contents if your file is not formatted nicely: Here’s how to for loading the file contents in such case: It takes a bit more effort if json object has some nested elements. This is one of…