It has been a long time since I last dealt with statistics, but I had an opportunity to touch on logistic regression at work—through spam filtering—so I did a little searching.
http://www.gen-info.osaka-u.ac.jp/testdocs/tomocom/cha.html#roji
is well organized and quite good. Moreover, it lets you perform regression on the page itself.
The expansion of the formulas is easier to follow here:
http://aoki2.si.gunma-u.ac.jp/lecture/Survival/mul-log.html
When it comes to spam filtering, Bayesian methods are all the rage, but of course logistic regression should also work. The question is how to choose the explanatory variables. That is probably where the know-how lies.
Related posts
Nikkei Sangyo Front Page: “Digital Arts Improves Browsing Restrictions; Patents in Japan and 22 Other Countries”???
Today’s Nikkei Sangyo front page has an article, “Digital Arts Improves Browsing Restrictions; Patents in Japan and 22 Other Countries.” According to the article, it seemed…
On the Symphonic Suite Space Battleship Yamato
The NHK Symphony Orchestra conducted by Naoto Otomo Composed by Kentaro Haneda Symphonic Suite Space Battleship Yamato https://youtu.be/lkkYEexkYVw?list=PLF1C1000D34C4616D The work was created using the various themes…
Behind the Scenes of Social Media Recommendations — Is Your Smartphone Eavesdropping on Your Conversations?! How Is Your Information Being Used Now? From NHK’s Chirushiru (5/31)
MyDataJapan board member Ota explains the dangerous digital realities you need to know Hello! Today, I would like to introduce some important points covered in NHK's…