Python inpの例

コード例 #1

ファイル: linear_classifier.py プロジェクト: lagercat/ml-systems

def main(argv):
    (train_x, train_y), (test_x, test_y) = load_data()

    feature_columns = [
        tf.feature_column.categorical_column_with_identity(key='Pclass',
                                                           num_buckets=4),
        tf.feature_column.categorical_column_with_vocabulary_list(
            key='Sex', vocabulary_list=['male', 'female']),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.categorical_column_with_vocabulary_list(
            key='Embarked', vocabulary_list=['C', 'Q', 'S']),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch')
    ]
    optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.001)

    model = tf.estimator.LinearClassifier(feature_columns=feature_columns,
                                          optimizer=optimizer)
    model.train(input_fn=lambda: inp(test_x, test_y), steps=10000)

    eval_result = model.evaluate(input_fn=lambda: inp(test_x, test_y))

    average_loss = eval_result['average_loss']
    print('\n' + 80 * '*')
    print('Error: ${:.0f}'.format(average_loss**0.5))
    print()

コード例 #2

def main(argv):

    (train_x, train_y), (test_x, test_y) = load_data()

    feature_columns = [
        tf.feature_column.categorical_column_with_identity(key='Pclass',
                                                           num_buckets=5),
        tf.feature_column.categorical_column_with_vocabulary_list(
            key='Sex', vocabulary_list=['male', 'female']),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.categorical_column_with_vocabulary_list(
            key='Embarked', vocabulary_list=['C', 'Q', 'S']),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch')
    ]

    optimizer = tf.train.AdagradOptimizer(learning_rate=0.1)
    model = tf.estimator.LinearClassifier(feature_columns=feature_columns,
                                          optimizer=optimizer)

    model.train(input_fn=lambda: inp(train_x, train_y, 'TRAIN'), steps=30000)

    eval_result = model.evaluate(input_fn=lambda: inp(test_x, test_y, 'EVAL'))
    average_loss = eval_result['average_loss']
    print('Average loss: ' + str(average_loss))

    brute_results = model.predict(
        input_fn=lambda: inp(load_submit(), (), 'PREDICT'))
    net_results = []
    for line in brute_results:
        net_results.append(line['class_ids'][0])
    write_to_file(net_results)

コード例 #3

ファイル: working_system.py プロジェクト: lagercat/ml-systems

def main(argv):

    feature_columns = [
        tf.feature_column.indicator_column(
            tf.feature_column.categorical_column_with_identity(
                key='Pclass', num_buckets=5)),
        tf.feature_column.indicator_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Sex',
                vocabulary_list=['male', 'female'])),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.indicator_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Embarked',
                vocabulary_list=['C', 'Q', 'S'])),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch'),
        tf.feature_column.numeric_column(key='agcl'),
        tf.feature_column.numeric_column(key='fsize'),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='title',
                vocabulary_list=['Mr', 'Mrs', 'Miss']),
            dimension=3),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='deck',
                vocabulary_list=cabin_list),
            dimension=3),
        tf.feature_column.numeric_column(key='Fare_Per_Person')
    ]

    (train_x, train_y), (test_x, test_y) = load_data(ratio=0.7)

    units = 2 * [30]
    optimizer = tf.train.AdagradOptimizer(learning_rate=0.1)
    model = tf.estimator.DNNClassifier(hidden_units=units,
                                       feature_columns=feature_columns,
                                       optimizer=optimizer,
                                       activation_fn=tf.nn.sigmoid)

    model.train(input_fn=lambda: inp(train_x, train_y, 'TRAIN', rep=3000),
                steps=600000)

    eval_result = model.evaluate(input_fn=lambda: inp(test_x, test_y,
                                 'EVAL', rep=1))
    average_loss = eval_result['average_loss']

    print('Average loss: ' + str(average_loss))

    brute_results = model.predict(input_fn=lambda: inp(load_submit(),
                                  (), 'PREDICT'))
    net_results = []
    for line in brute_results:
        net_results.append(line['class_ids'][0])
    write_to_file(net_results)

コード例 #4

ファイル: working_system.py プロジェクト: lagercat/ml-systems

def main(argv):

    feature_columns = [
        tf.feature_column.indicator_column(
            tf.feature_column.categorical_column_with_identity(key='Pclass',
                                                               num_buckets=5)),
        tf.feature_column.indicator_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Sex', vocabulary_list=['male', 'female'])),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.indicator_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Embarked', vocabulary_list=['C', 'Q', 'S'])),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch'),
        tf.feature_column.numeric_column(key='agcl'),
        tf.feature_column.numeric_column(key='fsize'),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='title', vocabulary_list=['Mr', 'Mrs', 'Miss']),
            dimension=3),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='deck', vocabulary_list=cabin_list),
            dimension=3),
        tf.feature_column.numeric_column(key='Fare_Per_Person')
    ]

    (train_x, train_y), (test_x, test_y) = load_data(ratio=0.7)

    units = 2 * [30]
    optimizer = tf.train.AdagradOptimizer(learning_rate=0.1)
    model = tf.estimator.DNNClassifier(hidden_units=units,
                                       feature_columns=feature_columns,
                                       optimizer=optimizer,
                                       activation_fn=tf.nn.sigmoid)

    model.train(input_fn=lambda: inp(train_x, train_y, 'TRAIN', rep=3000),
                steps=600000)

    eval_result = model.evaluate(
        input_fn=lambda: inp(test_x, test_y, 'EVAL', rep=1))
    average_loss = eval_result['average_loss']

    print('Average loss: ' + str(average_loss))

    brute_results = model.predict(
        input_fn=lambda: inp(load_submit(), (), 'PREDICT'))
    net_results = []
    for line in brute_results:
        net_results.append(line['class_ids'][0])
    write_to_file(net_results)

コード例 #5

ファイル: custom_estimator.py プロジェクト: lagercat/ml-systems

def main(argv):
    (train_x, train_y), (test_x, test_y) = load_data()

    feature_columns = [
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_identity(key='Pclass',
                                                               num_buckets=4),
            dimension=3),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Sex', vocabulary_list=['male', 'female']),
            dimension=3),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Embarked', vocabulary_list=['C', 'Q', 'S']),
            dimension=3),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch')
    ]

    model = tf.estimator.Estimator(model_fn=dnn_model_fn,
                                   params={
                                       'feature_columns': feature_columns,
                                       'learning_rate': 0.001,
                                       'optimizer':
                                       tf.train.GradientDescentOptimizer,
                                       'hidden_units': [20, 20]
                                   })
    model.train(input_fn=lambda: inp(train_x, train_y), steps=100)

コード例 #6

ファイル: custom_estimator.py プロジェクト: lagercat/ml-systems

def main(argv):
    (train_x, train_y), (test_x, test_y) = load_data()

    feature_columns = [
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_identity(
                key='Pclass', num_buckets=4),
            dimension=3),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Sex',
                vocabulary_list=['male', 'female']), dimension=3),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.embedding_column(
            tf.feature_column.categorical_column_with_vocabulary_list(
                key='Embarked',
                vocabulary_list=['C', 'Q', 'S']),
            dimension=3),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch')
    ]

    model = tf.estimator.Estimator(
            model_fn=dnn_model_fn,
            params={
                'feature_columns': feature_columns,
                'learning_rate': 0.001,
                'optimizer': tf.train.GradientDescentOptimizer,
                'hidden_units': [20, 20]
            }
    )
    model.train(input_fn=lambda: inp(train_x, train_y), steps=100)

コード例 #7

ファイル: linear_working_system.py プロジェクト: lagercat/ml-systems

def main(argv):

    (train_x, train_y), (test_x, test_y) = load_data()

    feature_columns = [
        tf.feature_column.categorical_column_with_identity(
            key='Pclass', num_buckets=5),
        tf.feature_column.categorical_column_with_vocabulary_list(
            key='Sex',
            vocabulary_list=['male', 'female']),
        tf.feature_column.numeric_column(key='Age'),
        tf.feature_column.numeric_column(key='Fare'),
        tf.feature_column.categorical_column_with_vocabulary_list(
            key='Embarked',
            vocabulary_list=['C', 'Q', 'S']),
        tf.feature_column.numeric_column(key='SibSp'),
        tf.feature_column.numeric_column(key='Parch')
    ]

    optimizer = tf.train.AdagradOptimizer(learning_rate=0.1)
    model = tf.estimator.LinearClassifier(feature_columns=feature_columns,
                                          optimizer=optimizer)

    model.train(input_fn=lambda: inp(train_x, train_y, 'TRAIN'),
                steps=30000)

    eval_result = model.evaluate(input_fn=lambda: inp(test_x, test_y,
                                 'EVAL'))
    average_loss = eval_result['average_loss']
    print('Average loss: ' + str(average_loss))

    brute_results = model.predict(input_fn=lambda: inp(load_submit(),
                                  (), 'PREDICT'))
    net_results = []
    for line in brute_results:
        net_results.append(line['class_ids'][0])
    write_to_file(net_results)