Ich versuche, TensorFlow für einen Datensatz mit einigen kategorialen Variablen zu verwenden. Ich habe sie mit Dummies kodiert, aber es sieht so aus, als ob es Probleme verursacht und TF beschwert sich, dass der Datensatz nicht dicht ist.Kategoriale Variablen in TensorFlow
Oder ist der Grund für den Fehler etwas völlig anderes?
Ich versuche, ein einfaches neuronales Netzwerkmodell mit 1 versteckter Schicht mit stochastischem Gradienten auszuführen. Der Code wurde zu arbeiten, wenn der Eingang numerischer Variablen (Bilder von Ziffern von MNIST)
dank
-------------------------------------------------------------------------- ValueError Traceback (most recent call last) in() 37 return(test_acc,round(l,5)) 38 ---> 39 define_batch(0.005) 40 run_batch()
in define_batch(beta) 11 shape=(batch_size, num_var)) 12 tf_train_labels = tf.placeholder(tf.float32, shape=(batch_size, num_labels)) ---> 13 tf_valid_dataset = tf.constant(valid_dataset) 14 tf_test_dataset = tf.constant(test_dataset) 15
/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/tensorflow/python/ops/constant_op.pyc in constant(value, dtype, shape, name) 159 tensor_value = attr_value_pb2.AttrValue() 160 tensor_value.tensor.CopyFrom( --> 161 tensor_util.make_tensor_proto(value, dtype=dtype, shape=shape)) 162 dtype_value = attr_value_pb2.AttrValue(type=tensor_value.tensor.dtype) 163 const_tensor = g.create_op(
/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/tensorflow/python/framework/tensor_util.pyc in make_tensor_proto(values, dtype, shape) 320 nparray = np.array(values, dtype=np_dt) 321 if list(nparray.shape) != _GetDenseDimensions(values): --> 322 raise ValueError("Argument must be a dense tensor: %s" % values) 323 # python/numpy default float type is float64. We prefer float32 instead. 324 if (nparray.dtype == np.float64) and dtype is None:
ValueError: Argument must be a dense tensor: Tuesday Wednesday Thursday Friday Saturday Sunday CENTRAL \ 736114
0.0 0.0 0.0 0.0 1.0 0.0 0.0 437148 0.0 0.0 1.0 0.0 0.0 0.0 0.0 605041 0.0 0.0 0.0 0.0 0.0 0.0 0.0 444608 0.0 0.0 0.0 0.0 1.0 0.0 0.0 695549 0.0 0.0 0.0 0.0 1.0 0.0 0.0 662807 0.0 0.0 0.0 1.0 0.0 0.0 0.0 238635 0.0 0.0 0.0 0.0 0.0 1.0 0.0 549524 0.0 0.0 0.0 1.0 0.0 0.0 0.0 705478 1.0 0.0 0.0 0.0 0.0 0.0 0.0 557716 0.0 0.0 0.0 1.0 0.0 0.0 0.0 41808 0.0 0.0 0.0 0.0 0.0 1.0 0.0 227235 1.0 0.0 0.0 0.0 0.0 0.0 0.0 848719 0.0 0.0 0.0 0.0 0.0 0.0 0.0 731202 0.0 0.0 0.0 0.0 1.0 0.0 0.0 467516 1.0 0.0 0.0 0.0 0.0 0.0 1.0
Hier ist ein Auszug aus dem Code
# Adding regularization to the 1 hidden layer network
def define_batch(beta):
batch_size = 128
num_RELU =256
graph1 = tf.Graph()
with graph1.as_default():
# Input data. For the training data, we use a placeholder that will be fed
# at run time with a training minibatch.
tf_train_dataset = tf.placeholder(tf.float32,
shape=(batch_size, num_var))
tf_train_labels = tf.placeholder(tf.float32, shape=(batch_size, num_labels))
tf_valid_dataset = tf.constant(valid_dataset)
tf_test_dataset = tf.constant(test_dataset)
# Variables.
weights_RELU = tf.Variable(
tf.truncated_normal([num_var, num_RELU]))
biases_RELU = tf.Variable(tf.zeros([num_RELU]))
weights_layer1 = tf.Variable(
tf.truncated_normal([num_RELU, num_labels]))
biases_layer1 = tf.Variable(tf.zeros([num_labels]))
# Training computation.
logits_RELU = tf.matmul(tf_train_dataset, weights_RELU) + biases_RELU
RELU_vec = tf.nn.relu(logits_RELU)
logits_layer = tf.matmul(RELU_vec, weights_layer1) + biases_layer1
# loss = tf.reduce_mean(
# tf.nn.softmax_cross_entropy_with_logits(logits_layer, tf_train_labels))
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(logits_layer, tf_train_labels,name="cross_entropy")
l2reg = tf.reduce_sum(tf.square(weights_RELU))+tf.reduce_sum(tf.square(weights_layer1))
beta = 0.005
loss = tf.reduce_mean(cross_entropy+beta*l2reg)
# Optimizer.
optimizer = tf.train.GradientDescentOptimizer(0.3).minimize(loss)
# Predictions for the training, validation, and test data.
train_prediction = tf.nn.softmax(logits_layer)
valid_prediction = tf.nn.softmax(
tf.matmul(tf.nn.relu((tf.matmul(tf_valid_dataset, weights_RELU) + biases_RELU)),weights_layer1)+biases_layer1)
test_prediction =tf.nn.softmax(
tf.matmul(tf.nn.relu((tf.matmul(tf_test_dataset, weights_RELU) + biases_RELU)),weights_layer1)+biases_layer1)
import datetime
startTime = datetime.datetime.now()
num_steps = 301 # change to 3001
def run_batch():
with tf.Session(graph=graph1) as session:
tf.initialize_all_variables().run()
print("Initialized")
for step in range(num_steps):
# Pick an offset within the training data, which has been randomized.
# Note: we could use better randomization across epochs.
offset = (step * batch_size) % (train_labels.shape[0] - batch_size)
# Generate a minibatch.
batch_data = train_dataset[offset:(offset + batch_size), :]
batch_labels = train_labels[offset:(offset + batch_size), :]
# Prepare a dictionary telling the session where to feed the minibatch.
# The key of the dictionary is the placeholder node of the graph to be fed,
# and the value is the numpy array to feed to it.
feed_dict = {tf_train_dataset : batch_data, tf_train_labels : batch_labels}
_, l, predictions, logits = session.run(
[optimizer, loss,train_prediction,logits_RELU], feed_dict=feed_dict)
if (step % 500 == 0):
print("Minibatch loss at step %d: %f" % (step, l))
print("Minibatch accuracy: %.1f%%" % accuracy(predictions, batch_labels))
print("Validation accuracy: %.1f%%" % accuracy(
valid_prediction.eval(), valid_labels))
test_acc = accuracy(test_prediction.eval(), test_labels)
print("Test accuracy: %.1f%%" % test_acc)
print('loss=%s' % l)
x = datetime.datetime.now() - startTime
print(x)
return(test_acc,round(l,5))
define_batch(0.005)
run_batch()
EDIT war: @gdhal Dank für es
train_dataset
suchen, ist ein Pandas Datenrahmen
train_dataset.columns
Index([u'Tuesday', u'Wednesday', u'Thursday', u'Friday', u'Saturday',
u'Sunday', u'CENTRAL', u'INGLESIDE', u'MISSION', u'NORTHERN', u'PARK',
u'RICHMOND', u'SOUTHERN', u'TARAVAL', u'TENDERLOIN', u' 3H - 4H',
u' 5H - 6H', u' 7H - 8H', u' 9H - 10H', u'11H - 12H', u'13H - 14H',
u'15H - 16H', u'17H - 18H', u'19H - 20H', u'21H - 22H', u'23H - 0H',
u'Xnorm', u'Ynorm', u'Hournorm'],
dtype='object')
Alle Variablen sind Dummies (mit 0 oder 1 Werten) mit Ausnahme der letzten 3 Variablen (Xnorm, Ynorm und Hournorm), bei denen es sich um auf das [0,1] Intervall normierte numerische Werte handelt. valid_dataset
test_dataset
und haben das gleiche Format
train_labels
eine Serie pandas ist
train_labels.describe()
count 790184
unique 39
top LARCENY/THEFT
freq 157434
Name: Category, dtype: object
valid_labels
und test_labels
haben das gleiche Format
Was geben Sie in den Platzhalter tf_train_dataset ein? – gdahl
@gdahl Ich habe den Beitrag bearbeitet und Informationen über die Daten hinzugefügt, die ich füttere. Vielen Dank –