def leave_one_out_error(self):
correct = 0
numPoints = len(self.full_set)
for i, x in enumerate(self.X):
print("LOO Progress:", i + 1, "/", numPoints)
known_label = self.labels[i]
beta = 0
if i == 0:
# Begining
beta = self.__optimize__(self.full_set[1:], self.labels[1:])
elif i == numPoints - 1:
# End
beta = self.__optimize__(self.full_set[:numPoints - 1],
self.labels[:numPoints - 1])
else:
# Middle
beta = self.__optimize__(
(Util.combineSets(self.full_set[:i],
self.full_set[i + 1:])),
Util.combineSets(self.labels[:i], self.labels[i + 1:]))
if self.classify(x, beta) == known_label:
correct += 1
incorrect = numPoints - correct
return 100 * (incorrect / numPoints)
def __init__(self, classA=False, classB=False, *, C=1, plotNow=False, printReport=False, brute_loo=True, brute_mute=False):
# Initialize Variables and Parameters
self.optimized = False
self.plotPrepared = False
self.constantC = C
self.supportVectors = []
self.margin = 0
self.printReport = printReport
self.brute = brute_loo
self.brute_mute = False
# Generate Random Data or Accept Provided Sets
if classA and classB:
self.classA = classA
self.classB = classB
self.fullSet = Util.combineSets(self.classA, self.classB)
self.fullSetX, self.fullSetY = self.fullSet.T
else:
self.classA, self.classB, self.fullSetX, self.fullSetY, self.fullSet = RandomData.random_data()
self.labels = RandomData.linear_labels()
# Dimension of Class Data
self.dimensions = len(self.classA[len(self.classA)-1])
# Optimize and Plot or Report if Needed
if plotNow or printReport:
self.optimize()
if plotNow:
print("Graphing SVM, Report Will Generate After Graph Is Closed If Requested.")
self.plot()
if printReport:
self.preparePlot()
print(self)