def NRMSE(train=True):
if train:
# 학습 데이터 사용
x = x_scaled_train
y_true = y_train
else:
# 테스트 데이터 사용
x = x_scaled_test
y_true = y_test
x = np.expand_dims(x, axis=0)
n = x.shape[1]
y_pred = load_model.predict(x)
y_pred_rescale = y_scaler.inverse_transform(y_pred[0])
RMSE = mean_squared_error(y_true, y_pred_rescale, squared=False)
print("RMSE: ", RMSE)
sum = 0
for i in range(n):
sum += y_true[i]
#sum = 실제 관측값 평균
sum /= n
return 100 * RMSE / sum
def plot_comparison(start_idx, length, train=True):
if train:
# 학습 데이터 사용
x = x_scaled_train
y_true = y_train
else:
# 테스트 데이터 사용
x = x_scaled_test
y_true = y_test
end_idx = start_idx + length
x = x[start_idx:end_idx]
x = np.expand_dims(x, axis=0)
y_true = y_true[start_idx:end_idx]
y_pred = load_model.predict(x)
y_pred_rescale = y_scaler.inverse_transform(y_pred[0])
signal_pred = y_pred_rescale[:, 0]
signal_true = y_true[:, 0]
plt.figure(figsize=(21, 7))
plt.plot(signal_true, label='true')
plt.plot(signal_pred, label='pred')
plt.ylabel('O3')
plt.legend()
plt.show()
def ValuePredictor(to_predict_list, size):
to_predict = np.array(to_predict_list).reshape(1, size)
if (size == 8): #Diabetes
loaded_model = pickle.load(
open('D:\AI\Project\End to end\Healthcare\Diabetes_model.pkl',
'rb'))
result = loaded_model.predict(to_predict)
if (size == 30): #Cancer
load_model = pickle.load(
open('D:\AI\Project\End to end\Healthcare\Cancer_model.pkl', 'rb'))
result = load_model.predict(to_predict)
elif (size == 10): #Liver
loaded_model = pickle.load(
open('D:\AI\Project\End to end\Healthcare\Liver_pickle.pkl', 'rb'))
result = loaded_model.predict(to_predict)
elif (size == 13): #Heart
loaded_model = pickle.load(
open('D:\AI\Project\End to end\Healthcare\Heart_pickle.pkl', 'rb'))
result = loaded_model.predict(to_predict)
return result[0]
with io.open(
"/content/drive/MyDrive/Final Project/New_Project/model/LSTM/tokenizer.json",
'w',
encoding='utf-8') as f:
f.write(json.dumps(tokenizer_json, ensure_ascii=False))
"""# Predict - LSTM"""
x_test
y_test
from tensorflow.keras.models import load_model
load_model = load_model('/content/best_LSTM_model.h5')
prediction = load_model.predict(X_test)
predictions = []
for i in tqdm(range(0, len(X_test))):
predictions.append(np.argmax(prediction[i]))
y_test = pd.DataFrame(y_test)
y_test[0:1]
prediction = model.predict(X_test)
np.argmax(prediction[1])
y_test1 = y_test.reindex().label
y_test1
"""# Load_Model
img = img / 255.0
"""**reshape into a single sample with 3 channels**"""
img = img.reshape(1, 32, 32, 3)
"""**predict the class**"""
predictions = load_model.predict_classes(img)
predictions
labels = ['airplane','automobile','bird','cat','deer','dog','frog','horse','ship','truck']
result = np.argmax(load_model.predict(img))
print(result)
labels[result]
"""**Test Image 2**"""
image_test= tf.keras.preprocessing.image.load_img('drive/My Drive/DATA files/CNN model weights/Deer.jpg', grayscale=False,
color_mode='rgb')
img_test3 = tf.keras.preprocessing.image.load_img('drive/My Drive/DATA files/CNN model weights/Deer.jpg',
grayscale=False,
color_mode='rgb',
target_size=(32,32))
img1 = np.array(img_test3)
img1.shape
def predict_tomorrow():
# 기상청 도시별 현재 날씨에서 수원의 기상 정보 받아옴
wheather_source = requests.get(
'https://www.weather.go.kr/weather/observation/currentweather.jsp')
soup = BeautifulSoup(wheather_source.content, "html.parser")
table = soup.find('table', {'class': 'table_develop3'})
current_data = np.zeros([1, 7])
flag = False
for tr in table.find_all('tr'):
tds = list(tr.find_all('td'))
for td in tds:
if td.find('a'):
point = td.find('a').text
if (point == '수원'):
current_data[0][0] = float(tds[5].text) #현재 기온(°C)
current_data[0][1] = float(
re.findall("\d+\.\d+",
tds[11].text.split('(')[1])[0]) #현재 풍속(m/s)
current_data[0][2] = float(tds[9].text) #현재 습도(%)
flag = True
break
if (flag):
break
# 공공데이터 포털의 한국환경공단_에어코리아_대기오염정보 api로 천천동 측정소의 대기 정보 받아옴
M = '&returnType=xml&numOfRows=100&pageNo=1&stationName=%EC%B2%9C%EC%B2%9C%EB%8F%99&dataTerm=DAILY&ver=1.0'
key = '0hZJZN2tb1cGI14XLG3tR7dATtivSKg9lF7f%2B58EEHvVnDGqxmxj5aXtarWm%2BNfKEPBxTMWKkduaYZLvdEBQOQ%3D%3D'
url = 'http://apis.data.go.kr/B552584/ArpltnInforInqireSvc/getMsrstnAcctoRltmMesureDnsty?serviceKey=' + key + M
response = requests.get(url)
soup2 = BeautifulSoup(response.content, "html.parser")
cv = soup2.find('item')
pm10 = soup2.find('pm10value').text
pm25 = soup2.find('pm25value').text
o3 = soup2.find('o3value').text
no2 = soup2.find('no2value').text
try:
current_data[0][3] = float(pm10) #현재 PM10(㎍/㎥)
current_data[0][4] = float(pm25) #현재 PM2.5(㎍/㎥)
current_data[0][5] = float(o3) #현재 오존농도(ppm)
current_data[0][6] = float(no2) #현재 이산화질소농도(ppm)
except ValueError:
print("현재 측정 값이 없습니다")
return
#print(current_data)
# 현재 시간 + 24시간 뒤의 오존 농도 예측
now = datetime.now()
tomorrow = now + timedelta(days=1) + timedelta(hours=9)
x_scaled_current = x_scaler.transform(current_data)
c = np.expand_dims(x_scaled_current, axis=0)
tomorrow_pred = load_model.predict(c)
tomorrow_pred_rescale = y_scaler.inverse_transform(tomorrow_pred[0])
O3_pred = float(str(tomorrow_pred_rescale[0])[2:-2])
O3_pred = round(O3_pred, 3)
year = tomorrow.year
month = tomorrow.month
day = tomorrow.day
hour = tomorrow.hour
print("수원시 장안구 천천동의 {0}년 {1}월 {2}일 {3}시 예상 오존 농도는 {4} ppm 입니다.".format(
year, month, day, hour, O3_pred))
font = cv2.FONT_HERSHEY_SIMPLEX
video_capture = cv2.VideoCapture(0)
while True:
# Capture frame-by-frame
ret, frame = video_capture.read()
frame = imutils.resize(frame, width=1000)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
faces = faceCascade.detectMultiScale(gray, 1.3, 5)
EMOTIONS_LIST = [
"Marah", "Jijik", "Takut", "Senang", "Netral", "Sedih", "Kaget"
]
for (x, y, w, h) in faces:
face_frame = gray[y:y + h, x:x + w]
face_frame = cv2.resize(face_frame, (48, 48))
preds = load_model.predict(face_frame[np.newaxis, :, :, np.newaxis])
label = EMOTIONS_LIST[np.argmax(preds)]
color = (0, 255, 0)
cv2.putText(frame, label, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 1,
color, 2)
cv2.rectangle(frame, (x, y), (x + w, y + h), color, 2)
# Display the resulting frame
cv2.imshow('Video', frame)
# out.write(frame)
if cv2.waitKey(33) & 0xFF == ord('q'):
break
video_capture.release()
# out.release()