pnn (version 1.0.1)

perf: Perf

Description

Performance of a Probabilist neural network.

Usage

perf(nn)

Arguments

nn

A trained and smoothed Probabilist neural network.

Value

A probabilist neural network updated with its performance.

Details

The function perf uses a hold-out method. This method takes the training set used by the function learn and iterate over each observation trying to guess the current observation with a reduced training set (without the current observation).It generates:

• Two lists of observed and guessed values.
• the following statistics: number of success and fails, a sucess rate (success_rate) and a bic indicator.

See Also

pnn-package, learn, smooth, guess, norms

Examples

library(pnn)
data(norms)
pnn <- learn(norms)
pnn <- smooth(pnn, sigma=0.8)
pnn <- perf(pnn)
pnn$observed
pnn$guessed
pnn$success
pnn$fails
pnn$success_rate
pnn$bic