tag:blogger.com,1999:blog-4318944459823143473.post8436182931731550905..comments2024-03-26T06:12:16.956+01:00Comments on Geeky is Awesome: Naive Bayes ClassificationUnknownnoreply@blogger.comBlogger2125tag:blogger.com,1999:blog-4318944459823143473.post-21798098707875606112016-04-19T21:10:21.101+02:002016-04-19T21:10:21.101+02:00You said that you know what they mean in a theoret...You said that you know what they mean in a theoretical sense, so I'll explain their difference in a practical sense.<br /><br />The difference is that P(like book | fairy=yes) can be found by counting how many books contain the word "fairy", whereas P(fairy=yes | like book) can be found by counting the number of books that you like. You also need to know how many of the books that you like contain "fairy", but that's a similarity not a difference.<br /><br />This might make you think that there is no point in changing one into the other since they can both be easily calculated. The problem is when you don't care only about the word "fairy" but also the word "gun". In this case P(like book | fairy=yes, gun=yes) is found by counting the number of books that contain both "fairy" and "gun", P(like book | fairy=yes, gun=no) is found by counting the number of books that contain "fairy" but not "gun", etc. whereas P(fairy=yes, gun=yes | like book) and all the other varieties are still found by just counting the number of books you like.<br /><br />Of course you still need to know how many of those books you like contain or don't contain "fairy" and "gun", but by using the Naive Bayes assumption we can simplify this to knowing how many books that you like contain "fairy" and how many books that you like contain "gun", rather than how many contain both "fairy" and "gun", how many contain "fairy" but not "gun", etc.mtantihttps://www.blogger.com/profile/10802419717511831691noreply@blogger.comtag:blogger.com,1999:blog-4318944459823143473.post-2970287094800270022016-04-10T09:27:23.775+02:002016-04-10T09:27:23.775+02:00Hello! I`m having a problem hero. You say that
&g...Hello! I`m having a problem hero. You say that <br />>which is another way of saying that a book with the word "fairy" has a 90% chance of being a good book.<br />So P(like | fairy = yes) = 90%<br />While at the bottom you say that <br />>Now we can use the table at the top to find P(word|Like book)<br />So my question is : what is the difference between P(like | fairy = yes) and P(fairy = yes | like). I know the definition of conditional probability etc. What i mean is i can`t find the difference in how you distinguish those 2.Oleg Stotskyhttps://www.blogger.com/profile/11135078528253948117noreply@blogger.com