Let assume I have a dataset like this dataset where there are several textual attributes even continuos attributes like age. I have always encountered cases where k-nn is applied on just two attributes and all of those ones numerical. How can I properly apply k-nn in a dataset like that?
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
2
You can code each of the categorical variables (such as race, marital status, etc.; the ones you are referring to as textual variables) as a set of boolean (1/0) dummy variables. So for instance you'd have one column for each race in the data set with a value of 1 if the subject was of that race and 0 otherwise, i.e for this data set you'd add 5 race columns, White, Asian-Pac-Islander, etc. So you'd add a You'd need to do that for each of the categorical variables.
See discussion here for more information on how to apply k-nn to problems like this.
-
$\begingroup$ After doing that which could be the best next move from your perspective in terms of calculating distance? $\endgroup$– MazzyCommented Jul 27, 2014 at 0:01
-
$\begingroup$ I don't think I can give you better advice than was given in the thread I pointed you too. I took a look at the thread and the most specific advice I saw for the categorical variables referred to Hamming distance for the binary variables. For the real-valued variables usually people use something Euclidean distance. So you'd need to do following: $\endgroup$– JeffMCommented Jul 27, 2014 at 16:46