def visualize_error(learner: str,
mode: str,
path='try_1/',
run=False,
block=100) -> bool:
"""visualizing error"""
manager = Mg.TestLearner(block_train=block)
if mode == 'binary':
labels = 1
bits = 2
elif mode == 'ternary':
labels = 1
bits = 3
elif mode == 'quaternary':
labels = 2
bits = 4
elif mode == 'sixteen':
labels = 2
bits = 16
else:
RuntimeError('mode=' + mode + ' is not defined!')
return False
file = mode + '/' + path
data = manager.get_error(learner=learner,
file=file,
bits=bits,
labels=labels,
run=run)
if learner == 'ml':
title = 'Maximum Likelihood'
elif learner == 'map':
title = 'Maximum A Posterior'
elif learner == 'bayes':
title = 'Bayes Classifier'
elif learner == 'logistic':
title = 'Logistic Regression'
elif learner == 'tree':
title = 'Decision Tree Classifier'
elif learner == 'vector':
title = 'Support Vector Machines Classifier'
elif learner == 'forest':
title = 'Random Forests Classifier'
elif learner == 'nnet':
title = 'Neural Networks Classifier'
else:
RuntimeError('learner=' + learner + ' is not defined')
return False
points, min_power, max_power = Mg.get_power()
plot = IO.Plotter(dark=True, x_title='1 / En (dB)', y_title='Error')
power_set = np.logspace(min_power, max_power, points)
power_set = [10 * np.log10(x_power) for x_power in power_set]
plot.to_log(x=power_set, y=data, color='blue', label=title)
plot.show()
return True
def test_output():
wb = xlwt.Workbook()
sheet1 = wb.add_sheet('Channel Use #' + str(1))
sheet1.write(0, 0, 'user id')
sheet1.write(0, 1, 'channel tap')
sheet1.write(0, 2, 'transmitter id')
sheet1.write(0, 3, 'channel')
sheet1.write(0, 4, 'time lot')
sheet1.write(0, 5, 'input')
sheet1.write(0, 6, 'desired')
sheet1.write(0, 7, 'output')
d = IO.DataWriter()
d.data_to_xl(wb, 'testing')
return
def __create_train_data(self, power: float, mode: str):
"""creating data set to train models"""
writer = IO.DataWriter()
gen = self.__model_train
if mode == 'binary':
data_set = gen.record_data_set_binary(power=power, factors=self.__factor, line=b_line)
elif mode == 'ternary':
data_set = gen.record_data_set_ternary(power=power, factors=self.__factor, line=t_line)
elif mode == 'quaternary':
data_set = gen.record_data_set_quadrature(power=power, factors=self.__factor, line=q_line)
else:
data_set = gen.record_data_set_sixteen(power=power, factors=self.__factor, line=s_line)
name = 'train_set'
# comment = self.__get_comment_train(power=power)
writer.frame_to_csv(data_set, name)
# writer.comment_csv(comment, name)
return
def __create_test_data(self, powers, mode: str):
"""creating data set to test models"""
writer = IO.DataWriter()
gen = self.__model_test
for power in powers:
if mode == 'binary':
data_set = gen.record_data_set_binary(power=power, factors=self.__factor, line=b_line)
elif mode == 'ternary':
data_set = gen.record_data_set_ternary(power=power, factors=self.__factor, line=t_line)
elif mode == 'quaternary':
data_set = gen.record_data_set_quadrature(power=power, factors=self.__factor, line=q_line)
else:
data_set = gen.record_data_set_sixteen(power=power, factors=self.__factor, line=s_line)
name = 'test_' + ('%.2f' % power)
# comment = self.__get_comment_test(power=power)
writer.frame_to_csv(data_set, name)
# writer.comment_csv(comment, name)
return
def __get_distribution_sixteen(path: str, file: str, rows: int, alpha: float,
labeled: bool) -> bool:
"""visualizing the distribution of data"""
data = 'sixteen/' + path + file
data_set = read_file(file=data, rows=rows)
x1 = data_set['out #1']
x2 = data_set['out #2']
plot = IO.Plotter(dark=False, x_title='bit #1', y_title='bit #2')
if labeled:
y1 = data_set['bit #1']
y2 = data_set['bit #2']
s00 = -3
s01 = -1
s10 = +1
s11 = +3
index_00 = np.where((y1 == s00) & (y2 == s00))[0]
index_01 = np.where((y1 == s00) & (y2 == s01))[0]
index_02 = np.where((y1 == s00) & (y2 == s10))[0]
index_03 = np.where((y1 == s00) & (y2 == s11))[0]
index_04 = np.where((y1 == s01) & (y2 == s00))[0]
index_05 = np.where((y1 == s01) & (y2 == s01))[0]
index_06 = np.where((y1 == s01) & (y2 == s10))[0]
index_07 = np.where((y1 == s01) & (y2 == s11))[0]
index_08 = np.where((y1 == s10) & (y2 == s00))[0]
index_09 = np.where((y1 == s10) & (y2 == s01))[0]
index_10 = np.where((y1 == s10) & (y2 == s10))[0]
index_11 = np.where((y1 == s10) & (y2 == s11))[0]
index_12 = np.where((y1 == s11) & (y2 == s00))[0]
index_13 = np.where((y1 == s11) & (y2 == s01))[0]
index_14 = np.where((y1 == s11) & (y2 == s10))[0]
index_15 = np.where((y1 == s11) & (y2 == s11))[0]
c = np.random.uniform(0, 1, size=(16, 3))
plot.to_scatter(x1[index_00],
x2[index_00],
edge_color=c[0],
label='0000',
marker='v',
alpha=alpha)
plot.to_scatter(x1[index_01],
x2[index_01],
edge_color=c[1],
label='0001',
marker='o',
alpha=alpha)
plot.to_scatter(x1[index_02],
x2[index_02],
edge_color=c[2],
label='0010',
marker='*',
alpha=alpha)
plot.to_scatter(x1[index_03],
x2[index_03],
edge_color=c[3],
label='0011',
marker='p',
alpha=alpha)
plot.to_scatter(x1[index_04],
x2[index_04],
edge_color=c[4],
label='0100',
marker='s',
alpha=alpha)
plot.to_scatter(x1[index_05],
x2[index_05],
edge_color=c[5],
label='0101',
marker='<',
alpha=alpha)
plot.to_scatter(x1[index_06],
x2[index_06],
edge_color=c[6],
label='0110',
marker='P',
alpha=alpha)
plot.to_scatter(x1[index_07],
x2[index_07],
edge_color=c[7],
label='0111',
marker='h',
alpha=alpha)
plot.to_scatter(x1[index_08],
x2[index_08],
edge_color=c[8],
label='1000',
marker='o',
alpha=alpha)
plot.to_scatter(x1[index_09],
x2[index_09],
edge_color=c[9],
label='1001',
marker='D',
alpha=alpha)
plot.to_scatter(x1[index_10],
x2[index_10],
edge_color=c[10],
label='1010',
marker='^',
alpha=alpha)
plot.to_scatter(x1[index_11],
x2[index_11],
edge_color=c[11],
label='1011',
marker='8',
alpha=alpha)
plot.to_scatter(x1[index_12],
x2[index_12],
edge_color=c[12],
label='1100',
marker='X',
alpha=alpha)
plot.to_scatter(x1[index_13],
x2[index_13],
edge_color=c[13],
label='1101',
marker='d',
alpha=alpha)
plot.to_scatter(x1[index_14],
x2[index_14],
edge_color=c[14],
label='1110',
marker='*',
alpha=alpha)
plot.to_scatter(x1[index_15],
x2[index_15],
edge_color=c[15],
label='1111',
marker='>',
alpha=alpha)
else:
plot.to_scatter(x1,
x2,
edge_color='black',
marker='o',
in_color='blue',
alpha=alpha)
plot.show()
return True
def read_file(file: str, rows=None):
"""reading file"""
reader = IO.DataReader()
data_set = reader.read_data(data_name=file, rows=rows)
return data_set
def get_distribution(mode='quaternary',
path='try_1/',
file='test_1.00',
rows=300,
alpha=1.0,
labeled=False) -> bool:
"""visualizing the distribution of data"""
if mode == 'sixteen':
return __get_distribution_sixteen(path=path,
file=file,
rows=rows,
alpha=alpha,
labeled=labeled)
if mode != 'quaternary':
RuntimeError('mode=' + mode + ' is not an option!')
return False
data = mode + '/' + path + file
data_set = read_file(file=data, rows=rows)
x1 = data_set['out #1']
x2 = data_set['out #2']
plot = IO.Plotter(dark=False, x_title='bit #1', y_title='bit #2')
if labeled:
y1 = data_set['bit #1']
y2 = data_set['bit #2']
s0 = -1
s1 = +1
index_0 = np.where((y1 == s0) & (y2 == s0))[0]
index_1 = np.where((y1 == s0) & (y2 == s1))[0]
index_2 = np.where((y1 == s1) & (y2 == s0))[0]
index_3 = np.where((y1 == s1) & (y2 == s1))[0]
plot.to_scatter(x1[index_0],
x2[index_0],
edge_color='purple',
label='00',
marker='o',
alpha=alpha)
plot.to_scatter(x1[index_1],
x2[index_1],
edge_color='red',
label='01',
marker='^',
alpha=alpha)
plot.to_scatter(x1[index_2],
x2[index_2],
edge_color='blue',
label='10',
marker='d',
alpha=alpha)
plot.to_scatter(x1[index_3],
x2[index_3],
edge_color='green',
label='11',
marker='s',
alpha=alpha)
else:
plot.to_scatter(x1,
x2,
edge_color='black',
marker='o',
alpha=alpha,
in_color='blue')
plot.show()
return True
def __init__(self, block_train=100):
"""initializing testing class"""
self.__reader = IO.DataReader()
self.__block_train = block_train
return