#
# Written by:
# -- 
# John L. Weatherwax                2009-04-21
# 
# email: wax@alum.mit.edu
# 
# Please send comments and especially bug reports to the
# above email address.
#
#-----

set.seed(0)

source('../Chapter18/restaurant_problem_gen_data.R')
source('../Chapter18/decision_tree_learning.R')

source('./perceptron_learning.R')

training_set_sizes = seq( 10, 100, by=10 )
tree_testing_set_accuracy = c()
perc_testing_set_accuracy = c()
for( tss in training_set_sizes ){
  mcs_tree = c()
  mcs_perc = c()
  for( mci in 1:10 ){
    if( mci!=1 ){
      print( sprintf("REST: Train set size %10d; mci= %5d; tree_accuracy= %5.3f; perc_accuracy= %5.3f", tss, mci, mean(mcs_tree), mean(mcs_perc) ) )
    }      

    # Get some training data from the restaurant problem (taken from chap_18_prob_11.R):
    # 
    D_train  = restaurant_problem_gen_data( tss )
    data     = D_train$data
    willwait = D_train$willwait

    # Create the initial attributes variable (an R list with items and keys as above):
    # Note our training set must have seen the possible values for each attribute.
    #
    attributes = list()
    for( cn in colnames(data) ){
      attributes[[cn]] = sort(as.matrix(unique(data[cn])))
    }

    # Compute the default classification: 
    #
    default = majority_vote(willwait) 

    # Build a model using the complete data and then use it to predict on the training data:
    # 
    tree = decision_tree_learning( data, willwait, attributes, default ) 

    # Learn a PERCEPTRON with a training set of the given size:
    #
    X = map_features_to_doubles( data )
    y = willwait + 0 # convert T/F to 0's and 1's 
    perceptron = perceptron_build_model( X, y )

    # Generate new data to test with:
    #
    D_test        = restaurant_problem_gen_data( tss )
    data_test     = D_test$data
    willwait_test = D_test$willwait

    yhat = decision_tree_predict_multiple_samples( tree, data_test ) 
    acc = sum(willwait_test == yhat)/tss
    mcs_tree = c(mcs_tree,acc)

    X = map_features_to_doubles( data_test ) 
    yhat = perceptron_predict( perceptron, X )
    acc = sum(willwait_test == yhat)/tss
    mcs_perc = c(mcs_perc,acc)
  }
  tree_testing_set_accuracy = c(tree_testing_set_accuracy, mean(mcs_tree))
  perc_testing_set_accuracy = c(perc_testing_set_accuracy, mean(mcs_perc))
}

#postscript("../../WriteUp/Graphics/Chapter20/chap_20_dup_fig_22_restaurant_problem.eps", onefile=FALSE, horizontal=FALSE)

plot( training_set_sizes, tree_testing_set_accuracy, col='red', type='l', xlab='training set size', ylab='Proporation correct on test set', ylim=c(0.4,1.0) )
lines( training_set_sizes, perc_testing_set_accuracy, col='green', type='l' )

grid()

legend( 10, 1.0, c('decision tree test accuracy','perceptron test accuracy'), lty=1, col=c('red','green') )

#dev.off()