Cross-validation for a Neural Network
cv.nn( object, K = 5, repeats = 1, decay = seq(0, 1, 0.2), size = 1:5, seed = 1234, trace = TRUE, fun, ... )
object | Object of type "nn" or "nnet" |
---|---|
K | Number of cross validation passes to use |
repeats | Repeated cross validation |
decay | Parameter decay |
size | Number of units (nodes) in the hidden layer |
seed | Random seed to use as the starting point |
trace | Print progress |
fun | Function to use for model evaluation (i.e., auc for classification and RMSE for regression) |
... | Additional arguments to be passed to 'fun' |
A data.frame sorted by the mean of the performance metric
See https://radiant-rstats.github.io/docs/model/nn.html for an example in Radiant
nn
to generate an initial model that can be passed to cv.nn
Rsq
to calculate an R-squared measure for a regression
RMSE
to calculate the Root Mean Squared Error for a regression
MAE
to calculate the Mean Absolute Error for a regression
auc
to calculate the area under the ROC curve for classification
profit
to calculate profits for classification at a cost/margin threshold
if (FALSE) { result <- nn(dvd, "buy", c("coupon", "purch", "last")) cv.nn(result, decay = seq(0, 1, .5), size = 1:2) cv.nn(result, decay = seq(0, 1, .5), size = 1:2, fun = profit, cost = 1, margin = 5) result <- nn(diamonds, "price", c("carat", "color", "clarity"), type = "regression") cv.nn(result, decay = seq(0, 1, .5), size = 1:2) cv.nn(result, decay = seq(0, 1, .5), size = 1:2, fun = Rsq) }