dsCondense
Data condensing
Contents
Syntax
- DS2=dsCondense(DS)
- DS2=dsCondense(DS, method)
- DS2=dsCondense(DS, method, plotOpt)
Description
DS2=dsCondense(DS) returns a reduced dataset using data condensing.
DS2=dsCondense(DS, method) uses the following methods to generate reduced dataset:
- method='mcs' for the method of minimal consistent set (See ref [2])
- method='random' for the method of random selection
- method='loo' for the method of leave-one-out.
DS2=dsCondense(DS, method, plotOpt) plots the result after data condensing (if the dimension of the dataset is 2)
References
- P. E. Hart, “The condensed nearest neighbor rule,” IEEE Trans. on Inform. Theory, vol. IT-14, pp. 515–516, May 1968.
- B. V. Dasarathy, "Minimal consistent set (MCS) identification for optimal nearest neighbor decision systems design," IEEE Trans. on Systems, Man, Cybernetics, vol. 24, pp. 511–517, Mar. 1994.
Example
DS=prData('peaks'); plotOpt=1; DS2=dsCondense(DS, 'mcs', plotOpt);
MCS data count = 1000 MCS data count = 96 MCS data count = 84 MCS data count = 83
