def linesLesen():
if (OSPATH.exists(LINES_FILE)):
file = open(LINES_FILE, "rt")
SCL.scale(file)
file.close()
else:
SYS.exit("[WARN]: File: " + LINES_FILE +
" konnte nicht gefunden werden!")
return
def __test__():
from bezier_generator import bezier_generate
from scaler import scale
from forest_generator import generate_forest
stack = []
points = []
class point :
def __init__(self, x, y) :
self.x = x
self.y = y
width = 90
height = 90
points.append(point(80 - randint(0, 60), 0))
points.append(point(40, 33))
points.append(point(60, 66))
points.append(point(width, 80 - randint(0, 60)))
obj = "="
gmap = bezier_generate(points, width, height, obj)
gmap = scale(gmap, "=", 2)
stack.append(gmap)
del points[3]
points.append(point(80 - randint(0, 60), height))
gmap = bezier_generate(points, width, height, obj)
stack.append(gmap)
gmap = scale(gmap, "=", 4)
stack = [merge(stack)]
trees = ["%", "$"]
trees_prop = [0.5, 0.5]
scale_value = 10
start_prop = 1.0
gmap = generate_forest(",", trees, trees_prop, width, height, scale_value, start_prop)
stack.append(gmap)
merge(stack)
for i in range(0, width) :
s = ""
for j in range(0, height) :
if (gmap[i][j] == None) :
s += "."
else :
s += gmap[i][j][-1]
print s
def machePolylineListe(points):
points = points + " "
polylinePointsX = list()
polylinePointsY = list()
polylineCMDs = list(("m"))
while len(points) != 0:
# Trennen der x und y Koordinaten der Polyline von den "points"
points = points.lstrip()
index = points.index(',')
polyNextX = float(points[0:index])
polylinePointsX.append(polyNextX)
points = points[index + 1:len(points)]
points = points.lstrip()
index = points.index(' ')
polyNextY = float(points[0:index])
polylinePointsY.append(polyNextY)
points = points[index + 1:len(points)]
points = points.lstrip()
polylineCMDs.append("l")
continue
polylineCMDs.pop()
SCALER.scale(polylinePointsX, polylinePointsY, polylineCMDs)
return
def machePolylineListe(points):
points = points + " "
polylinePointsX = list()
polylinePointsY = list()
polylineCMDs = list(("m"))
while len(points) != 0:
# separate xpolynext and ypolynext coord from points
points = points.lstrip()
index = points.index(',')
polyNextX = float(points[0:index])
polylinePointsX.append(polyNextX)
points = points[index + 1:len(points)]
points = points.lstrip()
index = points.index(' ')
polyNextY = float(points[0:index])
polylinePointsY.append(polyNextY)
points = points[index + 1:len(points)]
points = points.lstrip()
polylineCMDs.append("l")
continue
polylineCMDs.pop()
SCALER.scale(polylinePointsX, polylinePointsY, polylineCMDs)
return
def _preprocess(self, x):
assert self._training_finished, "Training is not finished"
x_scaled = x
# 1. Relace strings with numbers
x_scaled = model_util.replace_cat_variables(x=x_scaled,
catdict=self._catdict,
param_names=self._param_names)
# 2. Maybe encode data
if self._encode:
x_scaled, tmp_param_names = \
model_util.encode(sp=self._sp, x=x_scaled,
param_names=self._param_names,
num_folds=self._num_folds,
catdict=self._catdict)
assert self._used_param_names == tmp_param_names
# 3. Scale data
x_scaled = scaler.scale(scale_info=self._scale_info, x=x_scaled)
return x_scaled
def _set_and_preprocess(self, x, param_names):
assert not self._training_finished, "Training is not finished"
scaled_x = copy.deepcopy(x)
#if self._scale_info is None:
# raise ValueError("No scale info available for this model")
self.set_number_of_folds(scaled_x)
self._param_names = param_names
# 1. Replace strings with numbers
scaled_x = model_util.replace_cat_variables(scaled_x, self._catdict,
self._param_names)
# 2. Maybe encode data
if self._encode:
scaled_x, self._used_param_names = \
model_util.encode(sp=self._sp, x=scaled_x,
param_names=self._param_names,
num_folds=self._num_folds,
catdict=self._catdict)
else:
self._used_param_names = self._param_names
# 4. Set statistics
if self._encode:
#raise NotImplementedError("One-hot-encoding is not implemented")
self._scale_info = \
scaler.get_x_info_scaling_all(sp=self._sp, x=scaled_x,
param_names=self._used_param_names,
num_folds=self._num_folds,
encoded=self._encode)
else:
self._scale_info = \
scaler.get_x_info_scaling_all(sp=self._sp, x=scaled_x,
param_names=self._used_param_names,
num_folds=self._num_folds,
encoded=self._encode)
# 4. Scale data
scaled_x = scaler.scale(scale_info=self._scale_info, x=scaled_x)
return scaled_x
def machePathListe():
# print(punkteX)
# print(punkteY)
# print(punkteCmds)
SCALER.scale(punkteX, punkteY, punkteCmds)
return
#np.random.seed(1337) # for reproducibility
net_type = 'regression' # softmax|regression
mode = 'validate' # full_train|validate
activations = False
features_NN_prefix = net_type + '.'
# Scale
if not (os.path.exists(constants.train_features_scaled_file)
and os.path.exists(constants.test_features_scaled_file)):
var, mean = variance(constants.train_features_file)
sd = np.sqrt(var)
if not os.path.exists(constants.train_features_scaled_file):
print('Scaling ' + constants.train_features_scaled_file)
scale(constants.train_features_file,
constants.train_features_scaled_file, sd, mean)
if not os.path.exists(constants.test_features_scaled_file):
print('Scaling ' + constants.test_features_scaled_file)
scale(constants.test_features_file,
constants.test_features_scaled_file, sd, mean)
n_epochs = 200
n_models = 10
if mode == 'validate':
validate(n_epochs, n_models)
elif mode == 'full_train':
full_train(n_epochs, n_models)
else:
print('Unknown mode. Exiting.')
#np.random.seed(1337) # for reproducibility
net_type = 'regression' # softmax|regression
mode = 'validate' # full_train|validate
activations = False
features_NN_prefix = net_type + '.'
# Scale
if not (os.path.exists(constants.train_features_scaled_file) and
os.path.exists(constants.test_features_scaled_file)):
var, mean = variance(constants.train_features_file)
sd = np.sqrt(var)
if not os.path.exists(constants.train_features_scaled_file):
print('Scaling ' + constants.train_features_scaled_file)
scale(constants.train_features_file, constants.train_features_scaled_file, sd, mean)
if not os.path.exists(constants.test_features_scaled_file):
print('Scaling ' + constants.test_features_scaled_file)
scale(constants.test_features_file, constants.test_features_scaled_file, sd, mean)
n_epochs = 200
n_models = 10
if mode == 'validate':
validate(n_epochs, n_models)
elif mode == 'full_train':
full_train(n_epochs, n_models)
else:
print('Unknown mode. Exiting.')
