def convert_to_tuple(self, object, fields):
values = []
for field in fields:
value = object
for attr in field:
value = getvalue(value, attr)
values.append(value)
return tuple(values)
def apply(self, objects):
strings = []
for object in objects:
string = ""
for distinct in self:
string += getvalue(object, distinct.field, returnIfNone="")
if string in strings:
objects.remove(object)
else:
strings.append(string)
return objects
def apply(self, objects):
strings = []
for object in objects:
string = ''
for distinct in self:
string += getvalue(object, distinct.field, returnIfNone='')
if string in strings:
objects.remove(object)
else:
strings.append(string)
return objects
def get(self, obj, sensitive=True):
value = getvalue(obj, self.field_name, returnIfNone='')
if not sensitive:
return value.lower()
return value
def apply(self, objects):
return sorted(objects, key=lambda k: getvalue(k, self.field_name), reverse=self.reverse)
"lines.linewidth": 2,
'grid.linestyle': '--'
})
fpr, tpr, thre = roc_curve(y_test, prepro)
tpr_e = [tpr]
fpr_e = [fpr]
label_e = ['ROC curve (AUC = %0.4f)']
roc_e = [roc_auc]
roc_plot(tpr_e, fpr_e, label_e, roc_e)
###optimal threshold based on entire dataset
opt_thre.append(threshold_youden(fpr, tpr, thre))
###------threshold curves
(precision, recall, f1score, thresholds) = getvalue(y_test, prepro)
legend_e = ('precision', 'recall', 'f1score')
title_e = 'thresholds: entire dataset'
thresholds_e = [thresholds, thresholds, thresholds]
para_e = [precision, recall, f1score]
thre_plot(thresholds_e, para_e, legend_e, title_e)
print("\n")
print("\n")
'''
###age: XX~40, 41~60, 61~XX
group_test, age_pre, age_prepro, group_X = [[[] for _ in range(3)] for _ in range(4)]
for i in range(len(X_test)):
if X_test.loc[i, 'age'] < 41:
group_(age_pre[0], pre[i], group_test[0], y_test[i], age_prepro[0], prepro[i], group_X[0], X_test.loc[i])
elif X_test.loc[i, 'age'] < 71:
group_(age_pre[1], pre[i], group_test[1], y_test[i], age_prepro[1], prepro[i], group_X[1], X_test.loc[i])
def apply(self, objects):
return sorted(objects,
key=lambda k: getvalue(k, self.field_name),
reverse=self.reverse)
def control_car_process(data, status, run, start):
max_num = 2100
min_num = 700
turn_ratio = 337 #转弯率
turn_ratio_1 = 337 #一档转弯
turn_ratio_2 = 335 #二档转弯
turn_ratio_3 = 305 #三档转弯
turn_ratio_4 = 298 #四档转弯
sharp_turn = 800 #急转弯率
speed_offset = 20 #速度偏移量
while run.value:
speed_car = data[0]
angle_car = 1500
fn = '/dev/input/js0'
jsdev = open(fn, 'rb')
car = serial.value
axis_map, axis_states, button_map, button_states = getvalue()
lib_path = path + "/lib" + "/libart_driver.so"
so = cdll.LoadLibrary
lib = so(lib_path)
try:
if (lib.art_racecar_init(38400, car.encode("utf-8")) < 0):
raise
pass
lib.send_cmd(1500, 1500)
while run.value:
evbuf = jsdev.read(8)
if evbuf:
time, value, type, number = struct.unpack('IhBB', evbuf)
if type & 0x01:
button = button_map[number]
if button:
button_states[button] = value
#启动键设置为tl
if (button == "tl"
and button_states[button] == True):
start.value = True
print("START")
lib.send_cmd(speed_car, angle_car)
#停止按键
if ((button == "tr"
and button_states[button] == True)):
# Stop
print("Stop")
status.value = False
data[0] = 1500
data[1] = 1500
lib.send_cmd(1500, 1500)
#切换一档转向
if (button == "a"
and button_states[button] == True):
turn_ratio = turn_ratio_1
# 切换二档转向
if (button == "b"
and button_states[button] == True):
turn_ratio = turn_ratio_2
# 切换三档转向
if (button == "x"
and button_states[button] == True):
turn_ratio = turn_ratio_3
# 切换四档转向
if (button == "y"
and button_states[button] == True):
turn_ratio = turn_ratio_4
# 速度增加
if (button == "y"
and button_states[button] == True):
turn_ratio = turn_ratio_4
# 速度减小
if (button == "y"
and button_states[button] == True):
turn_ratio = turn_ratio_4
if (start.value == True
): # PS2 control speed and angle start
if type & 0x02:
axis = axis_map[number]
if axis:
#左滑轮x轴
if axis == "x":
fvalue = value / 32767
axis_states[axis] = fvalue
angle1 = 1500 - (fvalue * turn_ratio)
if angle1 <= min_num:
angle1 = min_num
if angle1 >= max_num:
angle1 = max_num
angle_car = int(angle1)
data[0] = speed_car
data[1] = angle_car
lib.send_cmd(speed_car, angle_car)
#右滑钮x轴，控制急转弯
if axis == "z":
fvalue = value / 32767
axis_states[axis] = fvalue
angle1 = 1500 - (fvalue * sharp_turn)
if angle1 <= min_num:
angle1 = min_num
if angle1 >= max_num:
angle1 = max_num
angle_car = int(angle1)
data[0] = speed_car
data[1] = angle_car
lib.send_cmd(speed_car, angle_car)
if axis == "rz": #右滑钮上下控制加速减速
fvalue = value / 32767
axis_states[axis] = fvalue
speed1 = args.speed - (fvalue *
speed_offset)
speed_car = int(speed1)
data[0] = speed_car
data[1] = angle_car
lib.send_cmd(speed_car, angle_car)
#左滑钮y轴为y，右滑钮y轴为rz
except:
print("car run error")
finally:
lib.send_cmd(1500, 1500)
print("car run finally")
def medication_processor(paths,
num_epochs,
num_embedding=500,
batch_size=256):
medication_info = {}
for path in paths:
with open(path, 'r') as fl:
js_data = ast.literal_eval(fl.read())
js_data = ast.literal_eval(js_data['data'].replace("'", "\""))
for data in js_data:
subject_id = data["subject"]
hospital_id = data["context"]
if subject_id not in medication_info:
medication_info[subject_id] = {}
if hospital_id not in medication_info[subject_id]:
medication_info[subject_id][hospital_id] = {}
drug = str(data["medicationCodeableConcept"]).lower()
quantity = "".join(data["quantity"].split())
quantity_value = getvalue(quantity)
# equating to string false because 0 is a valid value but is considered false
if quantity_value != "false":
quantity = quantity_value
elif len(re.findall("\d*\.?\d*-\d*\.?\d*", quantity)) > 0:
quantity = re.findall("\d*\.?\d*-\d*\.?\d*",
quantity)[0]
quantity = quantity.split("-")
quantity = list(map(getvalue, quantity))
quantity = sum(quantity) / len(quantity)
else:
quantity = 1
unit = ""
if data["note"] is not None:
unit = str(data["note"].split()[-1]).lower()
feature_name = drug + "_" + unit
if feature_name not in medication_info[subject_id][
hospital_id]:
medication_info[subject_id][hospital_id][
feature_name] = 0
medication_info[subject_id][hospital_id][
feature_name] += quantity
feature_dict = {}
row = []
columns = []
data = []
feature_index = 0
row_index = 0
medication_info_row_mapping = {}
for subject_id in medication_info:
medication_info_row_mapping[subject_id] = {}
for hospital_id in medication_info[subject_id]:
for feature in medication_info[subject_id][hospital_id]:
if feature not in feature_dict:
feature_dict[feature] = feature_index
feature_index = feature_index + 1
row.append(row_index)
columns.append(feature_dict[feature])
data.append(
medication_info[subject_id][hospital_id][feature])
medication_info_row_mapping[subject_id][
hospital_id] = row_index
row_index += 1
sparse_matrix = csr_matrix((data, (row, columns)),
shape=(row_index, feature_index))
inp_shape = sparse_matrix.shape[1]
textAutoEncoder = Autoencoder(inputshape=inp_shape,
encoding_dim=num_embedding)
autoencoder, encoder, decoder = textAutoEncoder.Model()
print("training model")
X_train, X_test = train_test_split(sparse_matrix,
test_size=0.33,
random_state=42)
autoencoder.fit(X_train,
X_train,
epochs=num_epochs,
batch_size=batch_size,
shuffle=True,
validation_data=(X_test, X_test),
verbose=False)
print("model fit finished")
medication_auto_encoder_data = {}
for subject_id in medication_info_row_mapping:
medication_auto_encoder_data[subject_id] = {}
for hospital_id in medication_info_row_mapping[subject_id]:
index = medication_info_row_mapping[subject_id][hospital_id]
encoded_ouput = encoder.predict(sparse_matrix[index])[0]
encoded_ouput = [float(a) for a in encoded_ouput]
medication_auto_encoder_data[subject_id][
hospital_id] = encoded_ouput
print("saving model data")
return medication_auto_encoder_data