def testReadFiles(self):
p = 0.5
eCsvReader = EgoCsvReader()
eCsvReader.setP(p)
dataDir = PathDefaults.getDataDir() + "infoDiffusion/"
egoFileName = dataDir + "EgoData3.csv"
alterFileName = dataDir + "AlterData10.csv"
examplesList, egoIndicesR, alterIndices, egoIndicesNR, alterIndicesNR = eCsvReader.readFiles(egoFileName, alterFileName)
#logging.debug(examplesList.getDataField("X"))
#Read in the ego and alter arrays
(egoArray, _) = eCsvReader.readFile(egoFileName, eCsvReader.getEgoQuestionIds())
(alterArray, _) = eCsvReader.readFile(alterFileName, eCsvReader.getAlterQuestionIds())
#Make up the correct results
numFeatures = examplesList.getDataFieldSize("X", 1)
numPersonFeatures = numFeatures/2
#Note: no alters in this case
numTransmissons = 6
X2 = numpy.zeros((numTransmissons, numFeatures))
y2 = numpy.zeros((numTransmissons, 1))
X2[0, 0:numPersonFeatures] = egoArray[0, :]
X2[0, numPersonFeatures:numFeatures] = egoArray[1, :]
y2[0, 0] = -1
X2[1, 0:numPersonFeatures] = egoArray[0, :]
X2[1, numPersonFeatures:numFeatures] = egoArray[2, :]
y2[1, 0] = -1
X2[2, 0:numPersonFeatures] = egoArray[1, :]
X2[2, numPersonFeatures:numFeatures] = egoArray[0, :]
y2[2, 0] = -1
X2[3, 0:numPersonFeatures] = egoArray[1, :]
X2[3, numPersonFeatures:numFeatures] = egoArray[2, :]
y2[3, 0] = -1
X2[4, 0:numPersonFeatures] = egoArray[2, :]
X2[4, numPersonFeatures:numFeatures] = egoArray[0, :]
y2[4, 0] = -1
X2[5, 0:numPersonFeatures] = egoArray[2, :]
X2[5, numPersonFeatures:numFeatures] = egoArray[1, :]
y2[5, 0] = -1
self.assertTrue((X2 == examplesList.getDataField("X")).all())
self.assertTrue((y2 == examplesList.getDataField("y")).all())
#Second test
#================
#I modified EgoData3 so that person 2 is the same age as person 1, and
# hence a homophile of 1. She (2) is excluded from the non-receivers, since
#she is a homophile of person 1.
p = 0
eCsvReader = EgoCsvReader()
eCsvReader.setP(p)
examplesList, egoIndicesR, alterIndices, egoIndicesNR, alterIndicesNR = eCsvReader.readFiles(egoFileName, alterFileName)
numTransmissons = 5
X2 = numpy.zeros((numTransmissons, numFeatures))
y2 = numpy.zeros((numTransmissons, 1))
X2[0, 0:numPersonFeatures] = egoArray[0, :]
X2[0, numPersonFeatures:numFeatures] = egoArray[2, :]
y2[0, 0] = -1
X2[1, 0:numPersonFeatures] = egoArray[1, :]
X2[1, numPersonFeatures:numFeatures] = egoArray[0, :]
y2[1, 0] = -1
X2[2, 0:numPersonFeatures] = egoArray[1, :]
X2[2, numPersonFeatures:numFeatures] = egoArray[2, :]
y2[2, 0] = -1
X2[3, 0:numPersonFeatures] = egoArray[2, :]
X2[3, numPersonFeatures:numFeatures] = egoArray[0, :]
y2[3, 0] = -1
X2[4, 0:numPersonFeatures] = egoArray[2, :]
X2[4, numPersonFeatures:numFeatures] = egoArray[1, :]
y2[4, 0] = -1
self.assertTrue((X2 == examplesList.getDataField("X")).all())
self.assertTrue((y2 == examplesList.getDataField("y")).all())
def testGenerateNonReceivers(self):
numEgos = 3
numFeatures = 5
p = 1
eCsvReader = EgoCsvReader()
eCsvReader.setP(p)
numContactsIndices = [0, 1]
homophileIndexPairs = [(2,3)]
receiverCounts = numpy.zeros(numEgos)
#First test a very simple example with 1 homophile pair
egoArray = numpy.zeros((numEgos, numFeatures))
egoArray[0, :] = [0, 1, 1, 5, 4]
egoArray[1, :] = [0, 1, 1, 5, 8]
egoArray[2, :] = [0, 0, 1, 3, 6]
(contactsArray, egoIndices, alterIndices) = eCsvReader.generateNonReceivers(egoArray, numContactsIndices, homophileIndexPairs, receiverCounts)
numContacts = 4
contactsArray2 = numpy.zeros((numContacts, numFeatures))
contactsArray2[0, :] = [0, 1, 1, 5, 8]
contactsArray2[1, :] = [0, 0, 1, 3, 6]
contactsArray2[2, :] = [0, 1, 1, 5, 4]
contactsArray2[3, :] = [0, 0, 1, 3, 6]
egoIndices2 = numpy.zeros(numContacts)
egoIndices2[0] = 0
egoIndices2[1] = 0
egoIndices2[2] = 1
egoIndices2[3] = 1
self.assertTrue((contactsArray == contactsArray2).all())
self.assertTrue((egoIndices == egoIndices2).all())
#Test the case when there are some receivers already
receiverCounts = numpy.array([1,1,1])
(contactsArray, egoIndices, alterIndices) = eCsvReader.generateNonReceivers(egoArray, numContactsIndices, homophileIndexPairs, receiverCounts)
numContacts = 2
contactsArray2 = numpy.zeros((numContacts, numFeatures))
contactsArray2[0, :] = [0, 1, 1, 5, 8]
contactsArray2[1, :] = [0, 1, 1, 5, 4]
egoIndices2 = numpy.zeros(numContacts)
egoIndices2[0] = 0
egoIndices2[1] = 1
self.assertTrue((contactsArray == contactsArray2).all())
self.assertTrue((egoIndices == egoIndices2).all())
#A more complex example
numEgos = 6
egoArray = numpy.zeros((numEgos, numFeatures))
egoArray[0, :] = [1, 1, 1, 5, 4]
egoArray[1, :] = [0, 0, 1, 5, 8]
egoArray[2, :] = [0, 0, 1, 3, 6]
egoArray[3, :] = [0, 0, 1, 5, 1]
egoArray[4, :] = [0, 0, 1, 5, 2]
egoArray[5, :] = [0, 0, 1, 5, 3]
receiverCounts = numpy.zeros(numEgos)
(contactsArray, egoIndices, alterIndices) = eCsvReader.generateNonReceivers(egoArray, numContactsIndices, homophileIndexPairs, receiverCounts)
numContacts = 4
contactsArray2 = numpy.zeros((numContacts, numFeatures))
contactsArray2[0, :] = [0, 0, 1, 5, 8]
contactsArray2[1, :] = [0, 0, 1, 5, 1]
contactsArray2[2, :] = [0, 0, 1, 5, 2]
contactsArray2[3, :] = [0, 0, 1, 5, 3]
egoIndices2 = numpy.zeros(numContacts)
egoIndices2[0] = 0
egoIndices2[1] = 0
egoIndices2[2] = 0
egoIndices2[3] = 0
self.assertTrue((contactsArray == contactsArray2).all())
self.assertTrue((egoIndices == egoIndices2).all())
#Test picking non-homophiles
egoArray[0, :] = [2, 1, 2, 5, 4]
(contactsArray, egoIndices, alterIndices) = eCsvReader.generateNonReceivers(egoArray, numContactsIndices, homophileIndexPairs, receiverCounts)
numContacts = 5
contactsArray2 = numpy.zeros((numContacts, numFeatures))
contactsArray2[0, :] = [0, 0, 1, 5, 8]
contactsArray2[1, :] = [0, 0, 1, 3, 6]
contactsArray2[2, :] = [0, 0, 1, 5, 1]
contactsArray2[3, :] = [0, 0, 1, 5, 2]
contactsArray2[4, :] = [0, 0, 1, 5, 3]
egoIndices2 = numpy.zeros(numContacts)
egoIndices2[0] = 0
egoIndices2[1] = 0
egoIndices2[2] = 0
egoIndices2[3] = 0
egoIndices2[4] = 0
self.assertTrue((contactsArray == contactsArray2).all())
self.assertTrue((egoIndices == egoIndices2).all())
#Choose different p
p = 0.5
eCsvReader.setP(p)
egoArray[0, :] = [1, 1, 1, 5, 4]
egoArray[1, :] = [0, 0, 1, 5, 8]
egoArray[2, :] = [0, 0, 1, 3, 6]
egoArray[3, :] = [0, 0, 1, 5, 8]
egoArray[4, :] = [0, 0, 1, 5, 8]
egoArray[5, :] = [0, 0, 1, 5, 8]
(contactsArray, egoIndices, alterIndices) = eCsvReader.generateNonReceivers(egoArray, numContactsIndices, homophileIndexPairs, receiverCounts)
numContacts = 3
contactsArray2 = numpy.zeros((numContacts, numFeatures))
contactsArray2[0, :] = [0, 0, 1, 5, 8]
contactsArray2[1, :] = [0, 0, 1, 5, 8]
contactsArray2[2, :] = [0, 0, 1, 3, 6]
egoIndices2 = numpy.zeros(numContacts)
egoIndices2[0] = 0
egoIndices2[1] = 0
egoIndices2[2] = 0
self.assertTrue((contactsArray == contactsArray2).all())
self.assertTrue((egoIndices == egoIndices2).all())
#Test 2 different homophile fields
p = 1
eCsvReader.setP(p)
numEgos = 6
numFeatures = 7
homophileIndexPairs = [(2,3), (4,5)]
egoArray = numpy.zeros((numEgos, numFeatures))
egoArray[0, :] = [0, 0, 1, 5, 1, 2, 1]
egoArray[1, :] = [0, 0, 1, 5, 1, 3, 2]
egoArray[2, :] = [0, 0, 1, 4, 1, 2, 3]
egoArray[3, :] = [1, 0, 1, 5, 1, 2, 4]
egoArray[4, :] = [0, 0, 1, 2, 1, 1, 5]
egoArray[5, :] = [0, 0, 1, 5, 1, 2, 6]
(contactsArray, egoIndices, alterIndices) = eCsvReader.generateNonReceivers(egoArray, numContactsIndices, homophileIndexPairs, receiverCounts)
numContacts = 2
contactsArray2 = numpy.zeros((numContacts, numFeatures))
contactsArray2[0, :] = [0, 0, 1, 5, 1, 2, 1]
contactsArray2[1, :] = [0, 0, 1, 5, 1, 2, 6]
egoIndices2 = numpy.zeros(numContacts)
egoIndices2[0] = 3
egoIndices2[1] = 3
self.assertTrue((contactsArray == contactsArray2).all())
self.assertTrue((egoIndices == egoIndices2).all())
