Description Usage Arguments Value Author(s) Examples
Train an elastic net logistic regression classifier
1 2 3 4 5 6 7 8 9 10 | trainGLR(truthLabels, predictors, lossMat, weight = rep(1, NROW(predictors)),
alphaVec = seq(0, 1, by = 0.2), tauVec = seq(0.1, 0.9, by = 0.05),
naFilter = 0.6, cvFolds = 5, seed = 1, verbose = FALSE)
## S3 method for class 'glmnetGLR'
print(glmnetGLRobject, ...)
## S3 method for class 'glmnetGLR'
predict(glmnetGLRobject, newdata, truthCol = NULL,
keepCols = NULL)
|
truthLabels |
the factor/character regression response variable |
predictors |
matrix whose columns are the explanatory regression variables |
lossMat |
a loss matrix specifying the penalties for classification errors |
weight |
the observation weights. The default value
is 1 for each observation. Refer to |
alphaVec |
The elastic net mixing parameter. Refer
to |
tauVec |
A sequence of tau threshold values for the logistic regression classifier. |
naFilter |
the proportion of data for a given predictor (column) that does not consist of missing data. If the proportion of sample observations which are NA > naFilter, then those predictors are removed from the regression analysis. |
cvFolds |
the number of cross validation folds |
seed |
the random seed for sampling during cross validation |
verbose |
set to |
glmnetGLRobject |
a train elastic net logistic classifier |
... |
Additional arguments to methods (ignored) |
glmnetGLRobject |
a trained elastic net logistic regression classifier |
newdata |
the new set of observations to be predicted. New data must be an array with the same column names as the training data |
... |
Additional arguments to predict method of
|
The elastic net logistic regression model that minimized
the expected loss over the set of alpha
,
lambda
, and tau
parameters.
Landon Sego
Alex Venzin
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 | # Load the VOrbitrap Shewanella QC data
data(traindata)
# Here we select the predictor variables
predictors <- as.matrix(traindata[,9:96])
# The logistic regression model requires a binary response
# variable.
resp <- traindata[,"response"]
# Specify the loss matrix. The "Poor" class is the target of interest.
# The penalty for misclassifying a "Poor" item as "Good" results in a
# loss of 5.
lM <- lossMatrix(c("Good","Good","Poor","Poor"),
c("Good","Poor","Good","Poor"),
c( 0, 1, 5, 0))
# Train the elastic net classifier
elasticNet <- trainGLR(truthLabels = resp,
predictors = predictors,
lossMat = lM)
# Observe the optimal alpha, lambda, and tau values that produced
# this elastic net logistic regression classifier.
print(elasticNet)
# Load the new observations
data(testdata)
# Use an elastic net regression classifier to make predictions about
# new observations for the response variable.
predict(elasticNet, testdata)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.