def main():
data = read_csv("term-deposit-marketing-2020.csv")
numeric_columns = ["age", "balance", "day", "duration", "campaign"]
binary_columns = ["default", "housing", "loan", "y"]
categorical_columns = ["job", "marital", "education", "contact", "month"]
'''no etiketli veri sayısı, yes etiketli veri sayısından çok fazla
olduğu için model yes etiketli verileri doğru tahmin edemiyor.
Bu problemi aşmak için imbalanced learning yöntemlerinden smote kullanıldı.'''
sm = SMOTE(sampling_strategy='auto', k_neighbors=2, random_state=0)
kf = KFold(n_splits=5, shuffle=True)
svc = SVC()
c = 13
acc_scores, rc_scores = modeling(data, c, kf, svc, numeric_columns,
binary_columns, categorical_columns, sm)
print(statistics.mean(acc_scores))
print(statistics.mean(rc_scores))
from preprocess import get_data, get_permutations, read_csv
from type_1_model import build_model
from type_1_loader import DataGenerator, get_valid
path = '/home/ubuntu/dataset/type1_train/{}'
# path = '/Users/lvyufeng/Documents/captcha_train_set/type1_train/{type1_train_19990.jpg}'
train_data, valid_data = read_csv(path.format('type1_train.csv'))
training_generator = DataGenerator(train_data, path)
x_valid, y_valid = get_valid(valid_data, path)
model = build_model()
model.summary()
model.fit_generator(training_generator,
epochs=3,
validation_data=(x_valid, y_valid),
max_queue_size=10,
workers=1)
super().train(features, targets)
self.model.fit(X=features, y=targets)
def accuracy_score(self, features, targets):
super().accuracy_score(features, targets)
score = self.model.score(features, targets, self.model.class_weight)
return score
def abs_errors(self, features, targets):
targets_pred = self.predict(features)
result = abs(targets_pred - targets)
return result
if __name__ == '__main__':
train_df = read_csv('../data/train_dataset.csv')
train_y = train_df.outcome
train_x = train_df.drop(columns=['outcome', 'date', 'team1', 'team2'])
test_df = read_csv('../data/matches_to_predict.csv')
test_y = test_df.outcome
test_x = test_df.drop(columns=['outcome', 'date', 'team1', 'team2'])
model = RandomForestModel(n_estimators=500, min_samples_split=5)
model.train(train_x, train_y)
model.save_model('../ckpts/random_forest_model.ckpt')
# model = RandomForestModel(n_estimators=100)
# model.load_model('../ckpts/random_forest_model.ckpt')
# score = model.accuracy_score(train_x, train_y)
"""
The main training script.
Written by Tanmay Patil
"""
import matplotlib.pyplot as plt
from model import create_model, train_model
from preprocess import object_to_date_time, get_daily_weather_data, normalize_data, read_csv, get_training_data
if __name__ == "__main__":
df = read_csv("../data/processed/delhi_weather_data_processed.csv")
df = object_to_date_time(df)
daily_weather = get_daily_weather_data(df)
daily_weather = normalize_data(daily_weather)
X,y = get_training_data(daily_weather)
X_train = X[:7300,::]
X_test = X[7300:,::]
y_train = y[:7300]
y_test = y[7300:]
model = create_model()
history = train_model(model, X_train, y_train)
train_loss = history.history['loss']
x_axis = [*range(1, len(train_loss + 1))]
plt.title('Training Loss')
plt.plot(x_axis, train_loss)
plt.show()
if __name__ == '__main__':
ckpt_dir = os.path.join(os.getcwd(), 'data', 'ckpts')
input_dir = os.path.join(os.getcwd(), 'data', 'input')
preprocess_dir = os.path.join(os.getcwd(), 'data', 'preprocess')
simulate_dir = os.path.join(os.getcwd(), 'data', 'simulate')
if not os.path.exists(ckpt_dir):
os.makedirs(ckpt_dir)
if not os.path.exists(input_dir):
os.makedirs(input_dir)
if not os.path.exists(preprocess_dir):
os.makedirs(preprocess_dir)
if not os.path.exists(simulate_dir):
os.makedirs(simulate_dir)
train_df = read_csv(os.path.join(preprocess_dir, 'train_dataset.csv'))
train_y = train_df.outcome
train_x = train_df.drop(columns=['outcome', 'date', 'team1', 'team2'])
test_df = read_csv(os.path.join(preprocess_dir, 'matches_to_predict.csv'))
test_y = test_df.outcome
test_x = test_df.drop(columns=['outcome', 'date', 'team1', 'team2'])
model = RandomForestModel(n_estimators=500, min_samples_split=5)
model.train(train_x, train_y)
model.save_model(os.path.join(ckpt_dir, 'random_forest_model.ckpt'))
# model = RandomForestModel(n_estimators=100)
# model.load_model('../ckpts/random_forest_model.ckpt')
# score = model.accuracy_score(train_x, train_y)