def test_training():
'''Launch a dummy training and ensure results repeatability'''
# Load preprocessed dataset
train_data, valid_data, classes = load_processed("data/processed/")
# Launch model training
train(train_data, valid_data, classes, output_dir="test/models/", n_iter=1)
# TODO: Compare files hash (depends on object stability)
pass
def run(cfg=None):
#Getting dataset
raw_data = data_acq(cfg)
#Building features
data = data_prep(raw_data, cfg)
#Split data
X_train, X_test, y_train, y_test = split_data(data)
#Hyperparameter optimization
opt = train(X_train, y_train, cfg)
#Evaluation report
generate_report(opt, X_test, y_test)
def start(self):
clean_data.clean()
build_features.build_features()
cv_scores = train_model.train()
self.cv_scores = cv_scores
import pandas as pd
from src.data import make_dataset
from src.features import build_features
from src.models import train_model
if __name__ == "__main__":
print("\n---- TEST FUNCTIONS -----\n")
# Get the dataset path
dataset_path = make_dataset.debug_path()
df = pd.read_csv(dataset_path, sep=',', low_memory=False)
print(f"Dataset: {df.head()}")
# Feature Engineering
X, y = build_features.preprocessing(df)
# Train the Model
dataset_splitted = train_model.split_dataset(X, y)
max_features = 'auto'
model = train_model.train(dataset_splitted, max_features)
import os
import pandas as pd
from src.features import preprocessing
from src.models import train_model
if __name__ == "__main__":
BASEPATH = os.path.abspath("data")
DATASET = "heart.csv"
dataset_path = os.path.join(BASEPATH, DATASET)
df = pd.read_csv(dataset_path, sep=',', low_memory=False)
# Feature Engineering
X, y = preprocessing.preprocessing(df)
# Split dataset and train model
dataset_splitted = train_model.split_dataset(X, y)
train_model.train(dataset_splitted)