def __init__(self, string, type_of_sentence=None):
self.type_of_sentence = type_of_sentence
self.sentence = string
self.nominal1 = None
self.nominal2 = None
self.get_nominal_pair()
self.remove_tags()
self.feature_set = FeatureSet(self)
def __init__(self, samplerate, featureList, mp3dirs, k, times, run_before, euclidean): self.cluster = [] self.SAMPLERATE = samplerate self.featureList = featureList self.num_features = len(featureList) self.mp3dirs = mp3dirs self.k = k self.times = times self.euclidean = euclidean #do mp3_to_feature_vectors and brand new FeatureSet object if this is a new dataset/feature combo if run_before == False: self.mp3_to_feature_vectors() self.f = FeatureSet(self.featureList, False) #else skip conversion, load from file else: self.f = FeatureSet(self.featureList, True)
class Sentence:
RelatorList = ["because", "after", "since", " as "]
def __init__(self, string, type_of_sentence=None):
self.type_of_sentence = type_of_sentence
self.sentence = string
self.nominal1 = None
self.nominal2 = None
self.get_nominal_pair()
self.remove_tags()
self.feature_set = FeatureSet(self)
# have to remove tags before tokenizing
def get_nominal_pair(self):
# finds the tags in the s and returns four cases
match = re.findall(r"(<.*?>(.*?)</.*?>)", self.sentence)
if match:
self.nominal1 = match[0][1]
self.nominal2 = match[1][1]
# returns the two cases we want which don't have the tags anymore
def remove_tags(self):
self.sentence = self.sentence.replace("<e>", "")
self.sentence = self.sentence.replace("</e>", "")
def is_causal_from_dataset(self):
return "Cause-Effect" in self.type_of_sentence
def causal_features(self):
features = {"HasRelator": self.feature_set.has_relator(),
"DiscreditRelator": self.feature_set.discredit_relator(),
"HasCausalVerb": self.feature_set.check_for_causal_verb(),
"HasCause": self.feature_set.check_for_cause_in_sentence()}
return features
from accurancyTools import *
from sklearn import ensemble
import itertools
import matplotlib.pyplot as plt
trainingDataSetNames = ['Volume', 'CentroidNorm', 'Centroid', 'Perimeter', 'PseudoRadius', 'Complexity',
'BoundingBox2Volume', 'BoundingBoxAspectRatio', 'IntensityMax', 'IntensityMean',
'IntensityMin', 'IntensityStd', 'CloseMassRatio', 'IntensityHist', 'gaussianCoefficients',
'gaussianGOV', 'Gradient', 'GradientOfMag']
senaryo1File = '../../noduledetectordata/ilastikoutput3/s1/s1.h5'
senaryo1LabelFile = '../../noduledetectordata/ilastikoutput3/s1/s1_labels.h5'
senaryo1BatchFile = '../../noduledetectordata/ilastikoutput3/s1/example_05.h5'
senaryo1BatchLabelFile = '../../noduledetectordata/ilastikoutput3/s1/labels_example_05.h5'
senaryo1 = {'all_features': senaryo1File, 'all_features_labels': senaryo1LabelFile, 'batch_features': senaryo1BatchFile, 'batch_features_labels': senaryo1BatchLabelFile}
senaryo1_train_sets = FeatureSet.readFromFile(senaryo1File, trainingDataSetNames, senaryo1LabelFile, 'labels')
senaryo1_batch_sets = FeatureSet.readFromFile(senaryo1BatchFile, trainingDataSetNames, senaryo1BatchLabelFile, 'labels')
senaryo1_sets = {'all_features_set': senaryo1_train_sets, 'batch_features_set': senaryo1_batch_sets, 'fileName': 'example05'}
senaryo2File = '../../noduledetectordata/ilastikoutput3/s2/s2.h5'
senaryo2LabelFile = '../../noduledetectordata/ilastikoutput3/s2/s2_labels.h5'
senaryo2BatchFile = '../../noduledetectordata/ilastikoutput3/s2/example_01.h5'
senaryo2BatchLabelFile = '../../noduledetectordata/ilastikoutput3/s2/labels_example_01.h5'
senaryo2 = {'all_features': senaryo2File, 'all_features_labels': senaryo2LabelFile, 'batch_features': senaryo2BatchFile, 'batch_features_labels': senaryo2BatchLabelFile}
senaryo2_train_sets = FeatureSet.readFromFile(senaryo2File, trainingDataSetNames, senaryo2LabelFile, 'labels')
senaryo2_batch_sets = FeatureSet.readFromFile(senaryo2BatchFile, trainingDataSetNames, senaryo2BatchLabelFile, 'labels')
senaryo2_sets = {'all_features_set': senaryo2_train_sets, 'batch_features_set': senaryo2_batch_sets, 'fileName': 'example01'}
senaryo3File = '../../noduledetectordata/ilastikoutput3/s3/s3.h5'
senaryo3LabelFile = '../../noduledetectordata/ilastikoutput3/s3/s3_labels.h5'
senaryo3BatchFile = '../../noduledetectordata/ilastikoutput3/s3/example_03.h5'
#'skewness']
#this is vigra computeed
#trainingDataSetNames = ['count', 'regionCenter', 'regionRadii', 'histogram']
# this is matlab
trainingDataSetNames = ['Volume', 'CentroidNorm', 'Perimeter','Complexity',
'BoundingBox2Volume','BoundingBoxAspectRatio', #, 'BoundingBoxAspectRatio',
'IntensityHist']#'BoundingBoxAspectRatio']#, 'IntensityMax','IntensityMean',
random_seed= 100
numpy.random.seed(random_seed)
print numpy.random.rand(1,1)
repeatN = 100
acc = numpy.zeros([repeatN, 3])
allFeatures = FeatureSet.readFromFile(featureFile, trainingDataSetNames,labelFile, 'labels')
coordinateFeatures = FeatureSet.readFromFile(featureFile,['Centroid'] ,labelFile, 'labels')
predThreshold = 0.5
thresholdRange = numpy.linspace(0.1, 0.999, 10)
roc_mn_acc=numpy.zeros([len(thresholdRange),3])
roc_std_acc=numpy.zeros([len(thresholdRange),3])
partreeCount = 50
counter = 0
for predThreshold in thresholdRange:
for ite in range(0,repeatN):
trn, val = allFeatures.divideSetRandom(1,1,True)
#trn, val = allFeatures.divideSetByZ(coordinateFeatures.data[:,2])
rf = vigra.learning.RandomForest(treeCount=partreeCount)
rf.learnRF(trn.data, trn.labels,randomSeed = ite*10)
'CloseMassRatio', 'IntensityHist', 'gaussianCoefficients', 'gaussianGOV',
'Gradient', 'GradientOfMag'
]
senaryo1File = '../../noduledetectordata/ilastikoutput3/s1/s1.h5'
senaryo1LabelFile = '../../noduledetectordata/ilastikoutput3/s1/s1_labels.h5'
senaryo1BatchFile = '../../noduledetectordata/ilastikoutput3/s1/example_05.h5'
senaryo1BatchLabelFile = '../../noduledetectordata/ilastikoutput3/s1/labels_example_05.h5'
senaryo1 = {
'all_features': senaryo1File,
'all_features_labels': senaryo1LabelFile,
'batch_features': senaryo1BatchFile,
'batch_features_labels': senaryo1BatchLabelFile
}
senaryo1_train_sets = FeatureSet.readFromFile(senaryo1File,
trainingDataSetNames,
senaryo1LabelFile, 'labels')
senaryo1_batch_sets = FeatureSet.readFromFile(senaryo1BatchFile,
trainingDataSetNames,
senaryo1BatchLabelFile, 'labels')
senaryo1_sets = {
'all_features_set': senaryo1_train_sets,
'batch_features_set': senaryo1_batch_sets,
'fileName': 'example05'
}
senaryo2File = '../../noduledetectordata/ilastikoutput3/s2/s2.h5'
senaryo2LabelFile = '../../noduledetectordata/ilastikoutput3/s2/s2_labels.h5'
senaryo2BatchFile = '../../noduledetectordata/ilastikoutput3/s2/example_01.h5'
senaryo2BatchLabelFile = '../../noduledetectordata/ilastikoutput3/s2/labels_example_01.h5'
senaryo2 = {
if __name__ == '__main__':
# test code
from Bio import GenBank
from Bio.SeqFeature import SeqFeature
from FeatureSet import FeatureSet
from GraphSet import GraphSet
from random import normalvariate
parser = GenBank.FeatureParser()
fhandle = open(
'/data/genomes/Bacteria/Nanoarchaeum_equitans/NC_005213.gbk', 'r')
genbank_entry = parser.parse(fhandle)
fhandle.close()
gdfs1 = FeatureSet(0, 'Nanoarchaeum equitans CDS - CDS')
gdfs2 = FeatureSet(1, 'Nanoarchaeum equitans CDS - gene')
for feature in genbank_entry.features:
if feature.type == 'CDS':
gdfs1.add_feature(feature)
if feature.type == 'gene':
gdfs2.add_feature(feature)
gdt = Track()
gdt.add_set(gdfs1)
gdt.add_set(gdfs2)
graphdata = []
for pos in xrange(1, len(genbank_entry.seq), 1000):
graphdata.append((pos, normalvariate(0.5, 0.1)))
gdgs = GraphSet(2, 'test data')
if __name__ == "__main__":
# test code
from Bio import GenBank
from Bio.SeqFeature import SeqFeature
from FeatureSet import FeatureSet
from GraphSet import GraphSet
from random import normalvariate
parser = GenBank.FeatureParser()
fhandle = open("/data/genomes/Bacteria/Nanoarchaeum_equitans/NC_005213.gbk", "r")
genbank_entry = parser.parse(fhandle)
fhandle.close()
gdfs1 = FeatureSet(0, "Nanoarchaeum equitans CDS - CDS")
gdfs2 = FeatureSet(1, "Nanoarchaeum equitans CDS - gene")
for feature in genbank_entry.features:
if feature.type == "CDS":
gdfs1.add_feature(feature)
if feature.type == "gene":
gdfs2.add_feature(feature)
gdt = Track()
gdt.add_set(gdfs1)
gdt.add_set(gdfs2)
graphdata = []
for pos in xrange(1, len(genbank_entry.seq), 1000):
graphdata.append((pos, normalvariate(0.5, 0.1)))
gdgs = GraphSet(2, "test data")
