gendata.mix: Generate binary data with Bayesian mixture models
Description
gendata.mix generates data (both training and test data) from
Bayesian mixture model. The prior distribution of "theta" is uniform(0,1). The value
of "alpha" is given by argument alpha, which controls the the overall
relationship between the response and the predictor variables.Usage
gendata.mix (n1,n2,m1,m2,p,alpha,prob.y=c(0.9,0.1))
Arguments
n1
the number of class 1 in training data
n2
the number of class 2 in training data
m1
the number of class 1 in test data
m2
the number of class 2 in test data
alpha
a parameter controlling the dependency between the features and the
response
prob.y
a vector of two elements specifying the probabilities of the response
being 1 in each group
Value
- train
- the training data, with the row standing for the cases and the
first column being the response
- test
- the test data, of the same format as "train"