def getMainFeatures(self, train, indexes, goal, samples=5000):
size = 300
np.random.seed(goal)
beginning = indexes[goal]
end = len(train) if goal == len(indexes) - 1 else indexes[goal + 1]
print str(samples) + " sequences of " + str(goal)
toChooseGoal = np.random.randint(beginning, end - size, size=samples)
print str(samples) + " sequences of other stuff than " + str(goal)
toChooseNotGoalRange = len(train) - size*(len(indexes)-1) - end + beginning
toChooseNotGoal = np.random.randint(0, toChooseNotGoalRange, size=samples)
self.translateArray(toChooseNotGoal, indexes, goal, size, end-beginning)
finalToChooseList = np.concatenate((toChooseGoal, toChooseNotGoal))
mapArray = np.array([train[i:i + size, :] for i in finalToChooseList])
print "Calculating stats for " + str(goal)
mainFeatures = getStats(mapArray)
print "Done with " + str(goal)
return mainFeatures
def getTestStats(self):
sequence = 0
indexSeq = 0
index = 0
print "Reading csv"
test = pd.read_csv('test.csv')
test = test.values
arrayTest = []
for element in range(len(test)):
if test[element][4] != sequence:
if sequence != 0:
print "Sequence "+str(index)
arrayTest.append(test[indexSeq:element])
indexSeq = element
index += 1
sequence = test[element][4]
print "Sequence "+str(index)
arrayTest.append(test[indexSeq:len(test)])
print "Calculating stats"
stats = getStats(np.array(arrayTest))
np.save("testStats", stats)
return stats
def getTestStats(self):
sequence = 0
indexSeq = 0
index = 0
print "Reading csv"
test = pd.read_csv('test.csv')
test = test.values
arrayTest = []
for element in range(len(test)):
if test[element][4] != sequence:
if sequence != 0:
print "Sequence " + str(index)
arrayTest.append(test[indexSeq:element])
indexSeq = element
index += 1
sequence = test[element][4]
print "Sequence " + str(index)
arrayTest.append(test[indexSeq:len(test)])
print "Calculating stats"
stats = getStats(np.array(arrayTest))
np.save("testStats", stats)
return stats