def main(unused_argv):
batchsize_video = 1
dir_path = os.path.dirname(os.path.realpath(__file__))
train_folder = os.path.join(dir_path, "train/")
test_folder = os.path.join(dir_path, "test/")
train_target = os.path.join(dir_path, 'train_target.csv')
my_solution_file = os.path.join(dir_path, 'solution.csv')
tf_record_dir = os.path.join(dir_path, 'tf_records')
os.makedirs(tf_record_dir, exist_ok=True)
tf_record_train = os.path.join(tf_record_dir, 'train' + '.tfrecords')
tf_record_test = os.path.join(tf_record_dir, 'test' + '.tfrecords')
if not os.path.exists(tf_record_train):
x_train = get_videos_from_folder(train_folder)
y_train = get_target_from_csv(train_target)
save_tf_record(x_train, tf_record_train, y=y_train)
if not os.path.exists(tf_record_test):
x_test = get_videos_from_folder(test_folder)
save_tf_record(x_test, tf_record_test)
# Create the Estimator
classifier = tf.estimator.Estimator(model_fn=cnn_model_fn,
model_dir="\\tmp\\model")
# Set up logging for predictions
# Log the values in the "Softmax" tensor with label "probabilities"
tensors_to_log = {"probabilities": "softmax_tensor"}
logging_hook = tf.train.LoggingTensorHook(tensors=tensors_to_log,
every_n_iter=50)
print('{}: Train'.format(datetime.now().strftime("%H:%M:%S")))
# Train the model
classifier.train(input_fn=lambda: input_fn_from_dataset(
tf_record_train, batch_size=batchsize_video),
max_steps=1,
hooks=[logging_hook])
print('{}: Evaluate'.format(datetime.now().strftime("%H:%M:%S")))
# Evaluate the model and print results
#eval_input_fn = tf.estimator.inputs.numpy_input_fn(x={"x": eval_data}, y=eval_labels, num_epochs=1, shuffle=False)
eval_results = classifier.evaluate(input_fn=lambda: input_fn_from_dataset(
tf_record_test, batch_size=batchsize_video))
print(eval_results)
print('{}: Predict'.format(datetime.now().strftime("%H:%M:%S")))
pred = classifier.predict(
input_fn=lambda: input_fn_from_dataset(tf_record_test,
batch_size=batchsize_video,
num_epochs=1,
shuffle=False))
print('{}: Save solution to {}'.format(datetime.now().strftime("%H:%M:%S"),
my_solution_file))
solution = prob_positive_class_from_prediction(pred)
save_solution(my_solution_file, solution)
from get_data import get_videos_from_folder, get_target_from_csv import os import numpy as np from utils import save_solution dir_path = os.path.dirname(os.path.realpath(__file__)) train_folder = os.path.join(dir_path, "train/") test_folder = os.path.join(dir_path, "test/") train_target = os.path.join(dir_path, 'train_target.csv') my_solution_file = os.path.join(dir_path, 'solution.csv') x_train = get_videos_from_folder(train_folder) y_train = get_target_from_csv(train_target) x_test = get_videos_from_folder(test_folder) dummy_solution = 0.1 * np.ones(len(x_test)) save_solution(my_solution_file, dummy_solution)
# Load datasets
print("# Loading data")
dir_path = "D:/Amr/OneDrive/_MyActivities/180922 ETH DAS Data Science/Advanced Machine Learning/Projects/Task4/"
train_folder = os.path.join(dir_path, "train/")
test_folder = os.path.join(dir_path, "test/")
# =============================================================================
# train_target = os.path.join(dir_path,'train_target.csv')
# my_solution_file = os.path.join(dir_path,'solution.csv')
# train_folder = "D:/Amr/OneDrive/_MyActivities/180922 ETH DAS Data Science/Advanced Machine Learning/Projects/Task4/train/"
# test_folder = "D:/Amr/OneDrive/_MyActivities/180922 ETH DAS Data Science/Advanced Machine Learning/Projects/Task4/test/"
# train_target = "D:/Amr/OneDrive/_MyActivities/180922 ETH DAS Data Science/Advanced Machine Learning/Projects/Task4/train_target.csv"
# #my_solution_file = os.path.join(dir_path,'solution.csv')
# =============================================================================
X_train_raw = get_videos_from_folder(train_folder)
X_test_raw = get_videos_from_folder(test_folder)
Id = pd.read_csv('solution.csv')['id']
y_train_raw = pd.read_csv('train_target.csv').drop(columns=['id'])
print("# Done")
print()
# Extract features from ECG waveforms
print("# Extracting features")
X_train_ft = ExtractECOFeatures(X_train_raw)
X_test_ft = ExtractECOFeatures(X_test_raw)
print("# Done")
print()
#%%
# Impute median values in empty cells