K - N N F P
                               ===========

K-NNFP (K-Nearest Neighbor on Feature Projections) is a
supervised inductive learning program. It is implemented in the
C language, and tested on Unix systems.

knnfp is invoked as

$ knnfp domain_name [-k k] [-v v]

Here, k : number of neighbors to be taken (default is 1),
      v : degree of the verbosity (default is 0).

Reads training data from <domain_name>.train,
and test data from <domain_name>.test.
Information about the dataset is read from <domain_name>.info.
Returns accuracy and run time in the file <domain_name>.res.

If verbosity option is set, then the intermediate activities
are reported to a file called <domain_name>.log.

As an example, for iris domain, call as

$ knnfp iris 

Make sure that files iris.info, iris.train and iris.test exist
in the current directory.

The knnfp reads information about the domain from the
<domain_name>.info file.
This file gives information about the number of features, their
types, and the class labes. Optionally, feature weights may be provided.
The line starting with the keyword "Features" gives information about
features. For example,

Features l l n l

indicates that there are 4 features; 1st, 2nd and 4th features
take on linear values, while the 3rd feature is nominal.
There are only 2 values are accepted; namely l (linear) 
and n (nominal).

Classes 1 2 4

indicates that there are 3 class labels; namely 1, 2 and 4.

An optional line starting with the keyword "Weights" provides 
the weights of the corresponding features. This line, if exists,
must come after the Features line.

Two general puropose c-shell scripts are provided for cross-validation
experiments.

cv <inducer> <domain> <fold> [<k>]
Experiments <inducer> on <domain> with <fold> fold cross validation.
Optional argument <k> is used as an argument for <inducer>

cvk=1-10 <inducer> <domain> <fold>
Experiments <inducer> on <domain> with <fold> fold cross validation
for K =1..10.