def process_data(self, classification_dict):
"""
Create and fill 3 list described in return section
:param classification_dict: filled dictionary with min, max, avg classification probabilities, counter and class value
:return: 3 list -> data (classification probabilities), Y (class values), metas (file names)
"""
metas, data, Y = [], [], []
for key, value in classification_dict.items():
class_value = None
tmp_data = []
for x, y in value.items():
tmp_data += [y[1], y[2], y[0] / float(y[3])]
class_value = y[-1]
if(class_value not in self.discrete_atributes):
self.discrete_atributes.append(class_value)
Y.append(self.discrete_atributes.index(class_value))
else:
Y.append(self.discrete_atributes.index(class_value))
metas.append([key])
data.append(tmp_data)
return numpy.array(data), numpy.array(Y), numpy.array(metas)
def lookup_from_function(class_var, bound, function):
"""
Construct ClassifierByDataTable or ClassifierByLookupTable
mirroring the given function.
"""
lookup = lookup_from_bound(class_var, bound)
if lookup:
for i, attrs in enumerate(Orange.utils.counters.LimitedCounter([len(var.values) for var in bound])):
lookup.lookup_table[i] = Orange.data.Value(class_var, function(attrs))
return lookup
else:
dom = Orange.data.Domain(bound, class_var)
data = Orange.data.Table(dom)
for attrs in Orange.utils.counters.LimitedCounter([len(var.values) for var in dom.features]):
data.append(Orange.data.Example(dom, attrs + [function(attrs)]))
return LookupLearner(data)
def lookup_from_function(class_var, bound, function):
"""
Construct ClassifierByDataTable or ClassifierByLookupTable
mirroring the given function.
"""
lookup = lookup_from_bound(class_var, bound)
if lookup:
for i, attrs in enumerate(
Orange.utils.counters.LimitedCounter(
[len(var.values) for var in bound])):
lookup.lookup_table[i] = Orange.data.Value(class_var,
function(attrs))
return lookup
else:
dom = Orange.data.Domain(bound, class_var)
data = Orange.data.Table(dom)
for attrs in Orange.utils.counters.LimitedCounter(
[len(var.values) for var in dom.features]):
data.append(Orange.data.Example(dom, attrs + [function(attrs)]))
return LookupLearner(data)
def _get_data(self, dir_path):
filecsv_list = []
for root, dirs, files in os.walk(dir_path):
for file in files:
if os.path.splitext(file)[1] == '.csv':
filecsv_list.append(os.path.join(root, file))
data = pd.DataFrame()
for csv in filecsv_list:
df_tmp = pd.read_csv(csv, header=0, encoding='utf-8')
data = data.append(df_tmp, ignore_index=True)
return data
if (firstLine != 1):
arr = []
i = 1
for val in line.split(";"):
if (i != 15):
arr.append(float(val))
elif (i == 15):
l = int(val)
leave.append(l)
if (l == 0):
countL0 += 1
else:
countL1 += 1
countObj += 1
i += 1
data.append(arr)
else:
firstLine = 0
P = countL0 / countObj
file = open(paramFile, "r")
alg = file.readline().rstrip()
if (alg == "H"):
Ncluster = int(file.readline())
metric = file.readline().rstrip()
link = file.readline().rstrip()
elif (alg == "K"):
Ncluster = int(file.readline())
initMeans = file.readline().rstrip()
algMeans = file.readline().rstrip()