def MDLPC_criterion(self, data, feature, cut_point):
'''
Determines whether a partition is accepted according to the MDLPC criterion
:param feature: feature of interest
:param cut_point: proposed cut_point
:param partition_index: index of the sample (dataframe partition) in the interval of interest
:return: True/False, whether to accept the partition
'''
#get dataframe only with desired attribute and class columns, and split by cut_point
data_partition = data.copy(deep=True)
data_left = data_partition[data_partition[feature] <= cut_point]
data_right = data_partition[data_partition[feature] > cut_point]
#compute information gain obtained when splitting data at cut_point
cut_point_gain = cut_point_information_gain(dataset=data_partition, cut_point=cut_point,
feature_label=feature, class_label=self._class_name)
#compute delta term in MDLPC criterion
N = len(data_partition) # number of examples in current partition
partition_entropy = entropy(data_partition[self._class_name])
k = len(data_partition[self._class_name].unique())
k_left = len(data_left[self._class_name].unique())
k_right = len(data_right[self._class_name].unique())
entropy_left = entropy(data_left[self._class_name]) # entropy of partition
entropy_right = entropy(data_right[self._class_name])
delta = log(3 ** k, 2) - (k * partition_entropy) + (k_left * entropy_left) + (k_right * entropy_right)
#to split or not to split
gain_threshold = (log(N - 1, 2) + delta) / N
if cut_point_gain > gain_threshold:
return True
else:
return False
def MDLPC_criterion(self, data, feature, cut_point):
'''
Determines whether a partition is accepted according to the MDLPC criterion
:param feature: feature of interest
:param cut_point: proposed cut_point
:param partition_index: index of the sample (dataframe partition) in the interval of interest
:return: True/False, whether to accept the partition
'''
#get dataframe only with desired attribute and class columns, and split by cut_point
data_partition = data.copy(deep=True)
data_left = data_partition[data_partition[feature] <= cut_point]
data_right = data_partition[data_partition[feature] > cut_point]
#compute information gain obtained when splitting data at cut_point
cut_point_gain = cut_point_information_gain(dataset=data_partition, cut_point=cut_point,
feature_label=feature, class_label=self._class_name)
#compute delta term in MDLPC criterion
N = len(data_partition) # number of examples in current partition
partition_entropy = entropy(data_partition[self._class_name])
k = len(data_partition[self._class_name].unique())
k_left = len(data_left[self._class_name].unique())
k_right = len(data_right[self._class_name].unique())
entropy_left = entropy(data_left[self._class_name]) # entropy of partition
entropy_right = entropy(data_right[self._class_name])
delta = log(3 ** k, 2) - (k * partition_entropy) + (k_left * entropy_left) + (k_right * entropy_right)
#to split or not to split
gain_threshold = (log(N - 1, 2) + delta) / N
if cut_point_gain > gain_threshold:
return True
else:
return False
def Gain(S, A):
gain = 0
values = {}
lenS = len(S)
Spos, Sneg = countPN(S, masterkey)
S_entropy = entropy(Spos, Sneg)
for s in S:
value = s[A]
if value not in values:
values[value] = [0, 0, 0]
if s[masterkey] == "yes":
values[value][0] = values[value][0] + 1
elif s[masterkey] == "no":
values[value][1] = values[value][1] + 1
acum = 0
S_At = 0
propt = 0
for k, v in values.items():
v[2] = entropy(v[0], v[1])
propt = (v[0] + v[1]) / lenS
acum = propt * v[2]
S_At = S_At - acum
return S_entropy + S_At
def compute_weighted_entropy(attribute_vals, ys):
""" Pivot by match value """
ys_by_match = defaultdict(list)
for is_match, y in izip(attribute_vals, ys):
ys_by_match[is_match].append(y)
""" Compute entropy, weighted by size of attribute """
wt_entropy = 0.0
num_items = float(len(ys))
for ismatch, lst in ys_by_match.items():
ent = entropy(lst)
wt = (len(lst) / num_items)
wt_entropy += wt * ent
return wt_entropy
def set_entropy(self):
tic = time()
node = self.root
queue = Queue()
queue.put(node)
while not queue.empty():
node = queue.get()
cnts = []
for child in node.children.values():
queue.put(child)
cnts.append(child.count)
if cnts:
cnts = np.array(cnts)
node.entropy = entropy(cnts)
else:
node.entropy = 0
])
testCol = ['EGITIM', 'YAS', 'CINSIYET']
testtData = pd.DataFrame(testManual, columns=testCol)
columns2 = ['EGITIM', 'YAS', 'CINSIYET', 'KABUL']
testDataManual = pd.DataFrame(testDataManual, columns=columns2)
# data = 'Qualitative_Bankruptcy.data.csv'
# cov19 = 'TimeAge.csv'
# df = pd.DataFrame(np.array(pd.read_csv(data)))
# columns = ['Industrial Risk', 'Management Risk', 'Financial Flexibility', 'Credibility', 'Competitiveness',
# 'Operating Risk', 'Class']
#
data = 'Qualitative_Bankruptcy.data.csv'
# # df = pd.DataFrame(pd.read_csv(data))
# # blueWins = df[['blueWins']]
# # df.drop(columns=['blueWins'],inplace=True)
# # df.insert(len(df.columns), 'blueWins', blueWins)
# # print(df.head())
e = entropy(testDataManual, 'karci', resultCol='KABUL', column_list=columns2)
e.calc()
e.result(testtData)
# for index, row in testtData.iterrows():
#
# print(row['YAS'])
# rangei =100
# rangej = 1
# for i in range(0,rangei):
# for j in range(0,rangej):
# print('\rData Loading {:.5f}% {} {}'.format((i+j/(rangej-1 if rangej != 1 else rangej))/(rangei)*100,i,j), end='\n', flush=False)