我正在尝试在tensorflow
中训练DNNClassifier
这是我的代码
train_input_fn = tf.estimator.inputs.pandas_input_fn(
x=X_train,
y=y_train,
batch_size=1000,
shuffle = True
)
nn_classifier = tf.estimator.DNNClassifier(hidden_units=[1300,1300,1300], feature_columns=X_train, n_classes=200)
nn_classifier.train(input_fn = train_input_fn, steps=2000)
这是y_train
的外观
[450 450 450 ... 327 327 327]
类型: numpy.ndarray
这就是X_train
的外观
[[ 9.79285 11.659035 1.279528 ... 1.258979 1.063923 -2.45522 ]
[ 8.711333 13.92955 1.117603 ... 3.588921 1.231256 -3.180302]
[ 5.159803 14.059619 1.740708 ... 0.28172 -0.506701 -1.326669]
...
[ 2.418473 0.542642 -3.658447 ... 4.631474 4.544892 -4.595605]
[ 6.51176 4.321688 -1.483697 ... 3.13299 5.476103 -2.833903]
[ 6.894113 5.986267 -1.178247 ... 2.305603 7.217919 -2.152574]]
类型: numpy.ndarray
错误:
in pandas_input_fn(x, y, batch_size, num_epochs, shuffle, queue_capacity, num_threads, target_column)
85 'Cannot use name %s for target column: DataFrame already has a '
86 'column with that name: %s' % (target_column, x.columns))
---> 87 if not np.array_equal(x.index, y.index):
88 raise ValueError('Index for x and y are mismatched.nIndex for x: %sn'
89 'Index for y: %sn' % (x.index, y.index))
更新1:使用numpy_input_fn
train_input_fn= tf.estimator.inputs.numpy_input_fn(
x=X_train,
y=y_train,
batch_size=1000,
shuffle = True
)
错误:
INFO:tensorflow:Calling model_fn.
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-23-3b7c6b879e38> in <module>()
10 start_time = time.time()
11 nn_classifier = tf.estimator.DNNClassifier(hidden_units=[1300,1300,1300], feature_columns=X_train, n_classes=200)
---> 12 nn_classifier.train(input_fn = train_input_fn, steps=2000)
13 total_time = start_time - time.time()
c:usersappdatalocalprogramspythonpython36libsite-packagestensorflowpythonestimatorestimator.py in train(self, input_fn, hooks, steps, max_steps, saving_listeners)
353
354 saving_listeners = _check_listeners_type(saving_listeners)
--> 355 loss = self._train_model(input_fn, hooks, saving_listeners)
356 logging.info('Loss for final step: %s.', loss)
357 return self
c:usersappdatalocalprogramspythonpython36libsite-packagestensorflowpythonestimatorestimator.py in _train_model(self, input_fn, hooks, saving_listeners)
822 worker_hooks.extend(input_hooks)
823 estimator_spec = self._call_model_fn(
--> 824 features, labels, model_fn_lib.ModeKeys.TRAIN, self.config)
825
826 if self._warm_start_settings:
c:usersappdatalocalprogramspythonpython36libsite-packagestensorflowpythonestimatorestimator.py in _call_model_fn(self, features, labels, mode, config)
803
804 logging.info('Calling model_fn.')
--> 805 model_fn_results = self._model_fn(features=features, **kwargs)
806 logging.info('Done calling model_fn.')
807
c:usersappdatalocalprogramspythonpython36libsite-packagestensorflowpythonestimatorcanneddnn.py in _model_fn(features, labels, mode, config)
347 head=head,
348 hidden_units=hidden_units,
--> 349 feature_columns=tuple(feature_columns or []),
350 optimizer=optimizer,
351 activation_fn=activation_fn,
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
有什么线索我在做什么错?
问题是估算器上的 feature_columns
参数。看一下tf.estimator.DNNClassifier
文档:
feature_columns
:一种含义的值,其中包含模型使用的所有特征列。集合中的所有项目都应是从_FeatureColumn
派生的类的实例。
文档中还有一个示例用法。您的X_train
看起来像许多数字列,在这种情况下,您可以简单地创建这样的列表:
feature_columns = [tf.feature_column.numeric_column(i) for i in range(...)]
我今天遇到了这个错误,并认为如果我证明了解决方案,那就太好了。问题是由tf.estimator.inputs.numpy_input_fn
带来的。根据TensorFlow文档,X
必须是pandas.DataFrame
实例,并且y
必须是pandas.Series
或pandas.DataFrame
实例。type()
功能可以帮助确定X_train
和y_train
值的数据类型。将X_train
和y_train
更改为适当的数据类型解决了问题。