def features1():
trainingSet, trainingComposers, numTrainingComposers, numTrainingPieces = loadData.loadTrainingFiles()
trainingFeatureVectors, trainingFeatureNames, trainingComposerNames = extractFeatures.initFeatureVectors(4, trainingSet, len(labels)+1)
testingSet, testingComposers, numTestingComposers, numTestingPieces = loadData.loadTestingFiles()
testingFeatureVectors, testingFeatureNames, testingComposerNames = extractFeatures.initFeatureVectors(4, testingSet, len(labels)+1)
nameList = list()
for name in trainingFeatureNames:
nameList.append(name)
for name in testingFeatureNames:
nameList.append(name)
print 'begin convertion'
trainingData = np.zeros((len(trainingFeatureVectors), len(nameList)))
testingData = np.zeros((len(testingFeatureVectors), len(nameList)))
for author in range(len(trainingFeatureVectors)):
for feature in trainingFeatureVectors[author]:
trainingData[author][nameList.index(feature)] = trainingFeatureVectors[author][feature]
for author in range(len(testingFeatureVectors)):
for feature in testingFeatureVectors[author]:
testingData[author][nameList.index(feature)] = testingFeatureVectors[author][feature]
print 'done convertion'
#data = np.array([[featureVectors[author][feature] for feature in sorted(featureVectors[author])] for author in range(len(featureVectors))])
predictions = bench_svm(trainingData, testingData, trainingComposerNames)
printStatistics(predictions, testingComposerNames)
def features2():
trainingSet, trainingComposers, numTrainingComposers, numTrainingPieces = loadData.loadTrainingFiles()
trainingData, trainingComposerNames = extractFeatures.initFeatureVectors(trainingSet, numTrainingComposers, numTrainingPieces)
testingSet, testingComposers, numTestingComposers, numTestingPieces = loadData.loadTestingFiles()
testingFeatureVectors, testingComposerNames = extractFeatures.initFeatureVectors(testingSet, numTestingComposers, numTestingPieces)
#trainingComposerNames = [0, 1, 2, 3, 4]
#testingComposerNames = [0, 1, 2, 3, 4]
predictions = bench_svm(trainingData, testingFeatureVectors, trainingComposerNames)
printStatistics(predictions, testingComposerNames, numTrainingComposers)
def features2():
#load data
dataSet, dataComposers, numTrainingComposers, numDataPieces = loadData.loadTrainingFiles()
data, composerNames = extractFeatures.initFeatureVectors(dataSet, numTrainingComposers, numDataPieces)
# split into a training and testing set
trainingSet, testingSet, trainingComposerNames, testingComposerNames = train_test_split(data, composerNames, test_size=0.20)
print 'training', len(trainingSet), len(trainingComposerNames), 'testing', len(testingSet), len(testingComposerNames)
#make predictions
predictions = bench_svm(trainingSet, testingSet, trainingComposerNames)
reducedTrainingSet, reducedTestingSet = dimReduction(2, trainingSet, testingSet, trainingComposerNames, testingComposerNames)
printStatistics(predictions, testingComposerNames)
plotData(reducedTrainingSet, trainingComposerNames, [0, 1, 2, 3, 4])