# X_train, X_test = seqRowOfLabelData[train_index], seqRowOfLabelData[test_index]
# y_train, y_test = labelVec[train_index], labelVec[test_index]
# #KNN
## neigh = KNeighborsClassifier(n_neighbors=7, metric = 'cosine', weights = 'distance')
# clf.fit(X_train, y_train)
# labelPredict = clf.predict(X_test)
# tmpAccuracyScoreList2.append(accuracy_score(y_test,labelPredict))
# sumarizeAccuracy2.append(np.average(tmpAccuracyScoreList2))
##
for k_KNN_feature_selection in range(5, 50): #k neightbor seq in knn graph
print k_KNN_feature_selection
fileName = "..//outputfile//" + version + "//" + testVer + "//knnGraph_" + str(
k_KNN_feature_selection) + ".npy"
sortedLaplaFeatureIndexes = np.array(
algFile.loadSortedLaplaFeatureIndexes(fileName))
# sortedFeatureRowMat = np.array([featureRowMatrix[index] for index in sortedLaplaFeatureIndexes])
sortedFeatureRowMat = np.array(
featureRowMatrix)[sortedLaplaFeatureIndexes]
sumarizeAccuracy = []
sumarizeAccuracy2 = []
#Laplacian score
for numOfFeature in kFeatures:
print numOfFeature
# build seqRowMat
seqRowMatrix = np.transpose(sortedFeatureRowMat[:numOfFeature])
seqRowOfLabelData = seqRowMatrix[:numOfLabelData]
# #build classifier
# neigh = KNeighborsClassifier(n_neighbors=11, metric = 'cosine', weights = 'distance')
for train_index, test_index in loo.split(seqRowOfLabelData):
X_train, X_test = seqRowOfLabelData[train_index], seqRowOfLabelData[test_index]
y_train, y_test = labelVec[train_index], labelVec[test_index]
#KNN
neigh = KNeighborsClassifier(n_neighbors=9, metric = 'cosine', weights = 'distance')
neigh.fit(X_train, y_train)
labelPredict = neigh.predict(X_test)
tmpAccuracyScoreList2.append(accuracy_score(y_test,labelPredict))
sumarizeAccuracy2.append(np.average(tmpAccuracyScoreList2))
for k_KNN_Graph_Preprocess in range(129,130): #k neightbor seq in knn graph
print k_KNN_Graph_Preprocess
fileName = "..//outputfile//"+version+ "//"+testVer+"//knnGraph_"+str(k_KNN_Graph_Preprocess) + ".npy"
sortedLaplaFeatureIndexes = np.array(algFile.loadSortedLaplaFeatureIndexes(fileName))
# sortedFeatureRowMat = np.array([featureRowMatrix[index] for index in sortedLaplaFeatureIndexes])
sortedFeatureRowMat = np.array(featureRowMatrix)[sortedLaplaFeatureIndexes]
sumarizeAccuracy = []
#Laplacian score
for numOfFeature in kFeatures:
# get seqRowMat
seqRowMatrix = np.transpose(sortedFeatureRowMat[:numOfFeature])
seqRowOfLabelData = seqRowMatrix[:numOfLabelData]
#leave one out
tmpAccuracyScoreList = []
for train_index, test_index in loo.split(seqRowOfLabelData):
from sklearn.feature_selection import VarianceThreshold
from sklearn.model_selection import KFold
import matplotlib.pyplot as plt
if __name__ == '__main__':
startTime = datetime.now()
fileh = tables.open_file("../outputfile/encodingFile.h5", mode="r")
featureRowMatrix = fileh.root.featureRowMatrix
sizeOfSeqList = featureRowMatrix.shape[1]
labelDict = algFile.readLabelDictFromFile()
labelVec = np.array(classifierFile.convertLabelDict2List(labelDict))
numOfLabelData = len(labelVec)
kFeatures = range(1, 175)
accuracyScoreList = []
accuracyScoreList2 = []
sortedLaplaFeatureIndexes = algFile.loadSortedLaplaFeatureIndexes()
sortedFeatureRowMat = np.array([featureRowMatrix[index] for index in sortedLaplaFeatureIndexes])
sumarizeAccuracy = []
sumarizeAccuracy2 = []
#variance
seqRowMatrixFull = algFile.readSeqRowMatFromFile()
sel = VarianceThreshold(threshold=(.8 * (1 - .8)))
newSeqRowMatrix = sel.fit_transform(seqRowMatrixFull)
newSeqRowOfLabelData = newSeqRowMatrix[:numOfLabelData]
kf = KFold(n_splits=20)
#Laplacian score
for numOfFeature in kFeatures:
print numOfFeature
from sklearn.feature_selection import VarianceThreshold
from sklearn.model_selection import KFold
import matplotlib.pyplot as plt
if __name__ == '__main__':
startTime = datetime.now()
fileh = tables.open_file("../outputfile/encodingFile.h5", mode="r")
featureRowMatrix = fileh.root.featureRowMatrix
sizeOfSeqList = featureRowMatrix.shape[1]
labelDict = algFile.readLabelDictFromFile()
labelVec = np.array(classifierFile.convertLabelDict2List(labelDict))
numOfLabelData = len(labelVec)
kFeatures = range(1, 200)
accuracyScoreList = []
accuracyScoreList2 = []
sortedLaplaFeatureIndexes = algFile.loadSortedLaplaFeatureIndexes(
"3_3.npy")
sortedFeatureRowMat = np.array(
[featureRowMatrix[index] for index in sortedLaplaFeatureIndexes])
sumarizeAccuracy = []
sumarizeAccuracy2 = []
#variance
seqRowMatrixFull = algFile.readSeqRowMatFromFile()
# sel = VarianceThreshold(threshold=(.6 * (1 - .6)))
sel = VarianceThreshold(threshold=0.2)
newSeqRowMatrix = sel.fit_transform(seqRowMatrixFull)
newSeqRowOfLabelData = newSeqRowMatrix[:numOfLabelData]
loo = LeaveOneOut()
#Laplacian score
for numOfFeature in kFeatures: