def CrossValidation(typeOfAutomata):
datasets = DataUtil.GenerateKFoldSet(dataPath + trainPl, dataPath + testPl)
results = []
for sets in datasets:
print("Making for new set")
tm = GetLearningAutomata(typeOfAutomata)
result = 0.0
TrainX = []
TrainY = []
for i in sets[0]:
TrainX.append(i[0])
TrainY.append(i[1])
TestX = []
TestY = []
for i in sets[1]:
TestX.append(i[0])
TestY.append(i[1])
TrainX = np.array(TsUtil.ReshapeData(TrainX, False))
TrainY = np.array(TrainY)
TestX = np.array(TsUtil.ReshapeData(TestX, False))
TestY = np.array(TestY)
inbetweenResults = []
tm.fit(TrainX, TrainY)
#result = 100*(tm.predict(np.array(TsUtil.ReshapeData(TestX, False))) == np.array(TestY)).mean()
result = tm.score(TestX, TestY) * 100
inbetweenResults.append(result)
print(" " + str(result))
results.append(inbetweenResults)
return results
def CrossValidation():
datasets = DataUtil.GenerateKFoldSet(dataPath + trainPl,dataPath + testPl)
results = []
for sets in datasets:
print("Making Tsetlin Machine for new set")
def GetMachine():
if convolutional:
return TM(clauses, T, s, (WindowX, WindowY), weighted_clauses=True, boost_true_positive_feedback=0)
else:
return TM(clauses, T, s, weighted_clauses=True, boost_true_positive_feedback=0)
tm = GetMachine()
result = 0.0
TrainX = []
TrainY = []
for i in sets[0]:
TrainX.append(i[0])
TrainY.append(i[1])
TestX = []
TestY = []
for i in sets[1]:
TestX.append(i[0])
TestY.append(i[1])
inbetweenResults = []
for i in range(epochs):
tm.fit(np.array(TsUtil.ReshapeData(TrainX, convolutional)),np.array(TrainY),epochs=1,incremental=True)
result = 100*(tm.predict(np.array(TsUtil.ReshapeData(TestX, convolutional))) == np.array(TestY)).mean()
inbetweenResults.append(result)
print(" " + str(result))
results.append(inbetweenResults)
return results
