classification experiments with the k-Nearest Neighbor algorithm.

Hi everyone

It's my first effort for KNN classification and am using wine data sample from UCI website. i was wondering what distance measures I can use in this case and how to get the computation time required to compute the distance from a sample to each other sample.

I want to determine its k nearest neighbors for each labeled sample, if the label of the sample is more common than any other label for the k nearest neighbors, then count the classification as correct. Otherwise, count the classification as incorrect.

I'd appreciate any input

Thanks

data set info: attributes :
      1) Alcohol
  2) Malic acid
  3) Ash
  4) Alcalinity of ash  
and ... 13)xxx

it categorizes wines into 3 classes

one example would be:

1,14.23,1.71,2.43,15.6,127,2.8,3.06,.28,2.29,5.64,1.04,3.92,1065

[Walter Roberson]

Measuring computation time is difficult to do right, especially if what you are trying to do is predict how efficient different methods would be with larger datasets.

[Husky]

Thanks walter, this dataset is small, only 178 lines each having 13 attributes

Answer 1

[Walter Roberson]

For timing, see the File Exchange Contribution timeit

[Husky]

Thanks but how exactly should I use this function in KNN code?