Leave-one-out One-nearest-neighbor Evaluation

Evaluates a clustering distance matrix within a supervised learning scheme: leave-one-out one-neares-neighbor cross-validation. This yields a rough estimate of the suitabilty of a distance function for discriminating between classes if ground-truth is known.

loo1nn(x, y)

Arguments

x

A pdclust object.

y

A vector of the true class labels.

Value

Returns a percentage-correct estimate.

References

Brandmaier, A. M. (2015). pdc: An R Package for Complexity-Based Clustering of Time Series. Journal of Statistical Software, 67(5), 1--23.

Author

Andreas Brandmaier brandmaier@mpib-berlin.mpg.de

See also

pdc.dist pdclust