def randomseries(n):
'''
Gerador de Série Temporal Estocástica - V.1.2 por R.R.Rosa
Trata-se de um gerador randômico não-gaussiano sem classe de universalidade via PDF.
Input: n=número de pontos da série
res: resolução
'''
res = n/12
df = pd.DataFrame(np.random.randn(n) * np.sqrt(res) * np.sqrt(1 / 128.)).cumsum()
a=df[0].tolist()
a=funcs.normalize(a)
x=range(0,n)
return x,a
from sklearn.preprocessing import LabelEncoder
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
import funcs
data = pd.read_csv('IoT.csv', delimiter=',')
labels = pd.DataFrame(data, columns=['label'])
data = pd.DataFrame(data, columns=['duration', 'orig_bytes', 'orig_pkts', 'proto', 'resp_bytes', 'conn_state',
'resp_pkts'])
quantity = ['duration', 'orig_bytes', 'orig_pkts', 'resp_bytes', 'resp_pkts']
data = funcs.normalize(data, quantity)
data = pd.get_dummies(data, columns=['proto', 'conn_state'])
dtype = float;
# print(data)
trust_old = dict()
for index in quantity:
trust_old[index] = [np.mean(data[index]) - (3 * np.std(data[index])),
np.mean(data[index]) + (3 * np.std(data[index]))]
data.loc[(data[index] > trust_old[index][1]) | (data[index] < trust_old[index][0]), index] = np.nan
data = data.dropna()
le = LabelEncoder()
testSample = pd.read_csv(baseLoc + 'dataManagerFiles/train/' +
"testwithCovars.csv",
index_col=pk)
testSample.columns = [
name.replace("testWithDummies", "mainWithDummies")
for name in testSample.columns
]
main = main.fillna(main.mean())
# maxi=main.max()
# mini=main.min()
# diff=pd.DataFrame(index=main.coulmns,values=maxi-mini)
# zeroes=diff[diff[0]==0].columns
testSample = testSample.fillna(main.mean())
main, testSample = normalize(main.drop(target, axis=1), testSample)
nanColumns = main.columns[main.isna().any()].tolist()
main = main.drop(nanColumns, axis=1)
testSample = testSample.drop(nanColumns, axis=1)
main = main.join(tar)
test = main.sample(n=int(main.shape[0] * 0.20), random_state=0)
train = main.drop(test.index, axis=0)
test_y = test[[target]]
train_y = train[[target]]
train = train.drop(target, axis=1)
test = test.drop(target, axis=1)
X = train
y = train_y
varSelected = [
'Age0mainWithDummies', 'TicketMeanT0mainWithDummies',
validation = 0
if validation == 1:
test = train.sample(n=10000)
train = train.drop(test.index, axis=0)
train_y = train['TARGET']
test_y = test['TARGET']
varSelected = [
'EXT_SOURCE_3', 'EXT_SOURCE_2', 'EXT_SOURCE_1', 'DAYS_EMPLOYED',
'AMT_GOODS_PRICE', 'DAYS_CREDIT_min', 'PRODUCT_d_mean',
'REGION_RATING_CLIENT_W_CITY', 'lowOccupation', 'utilization_-6_mean_max',
'payTominDue_-6_mean_mean', 'EXT_SOURCE_3', 'EXT_SOURCE_2', 'EXT_SOURCE_1',
'utilization_-6_mean_max', 'payTominDue_-6_mean_mean'
] #,'active_sumbur','active_mean'
#train,test=normalize(train[varSelected],test[varSelected])
train, test = normalize(train[varSelected], test[varSelected])
train['TARGET'] = train_y
test['TARGET'] = test_y
train.describe().to_csv(
"/home/pooja/PycharmProjects/datanalysis/finalDatasets/des.csv")
abc = AdaBoostClassifier(n_estimators=200, learning_rate=1)
# Train Adaboost Classifer
mlp = abc.fit(train[varSelected], train['TARGET'])
#Predict the response for test dataset
trainer = pd.DataFrame(mlp.predict_proba(train[varSelected].values),
columns=['good', 'TARGET'],
index=train.index)[['TARGET']]
submision = pd.DataFrame(mlp.predict_proba(test[varSelected].values),
columns=['good', 'TARGET'],
face_detector = mtcnn.MTCNN()
face_encoder = load_model(encoder_model)
encoding_dict = dict()
for person_name in os.listdir(people_dir):
person_dir = os.path.join(people_dir, person_name)
encodes = []
for img_name in os.listdir(person_dir):
img_path = os.path.join(person_dir, img_name)
img = cv2.imread(img_path)
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
results = face_detector.detect_faces(img_rgb)
if results:
box = max(results, key=lambda b: b['box'][2] * b['box'][3])
l_e = box['keypoints']['left_eye']
r_e = box['keypoints']['right_eye']
face = align(img, l_e, r_e, size=required_size, eye_pos=(0.35, 0.4))
face = normalize(face)
encode = get_encode(face_encoder, face, required_size)
encodes.append(encode)
if encodes:
encode = np.sum(encodes, axis=0)
encode = l2_normalizer.transform(encode.reshape(1, -1))
encoding_dict[person_name] = encode[0]
for key in encoding_dict.keys():
print(key)
with open(encodings_path, 'bw') as file:
pickle.dump(encoding_dict, file)