我试图在具有一些分类变量的数据集上使用TensorFlow。我用傻瓜编码了它们，但看起来它引起了麻烦，TF抱怨数据集不密集。
或者错误的原因是什么？
我试图运行一个简单的神经网络模型，其中1个隐藏层具有随机梯度。当输入是数字变量（来自MNIST的数字图像）
时，代码正在工作感谢
----------------------------------------------- --------------------------- ValueError Traceback（最近一次调用 最后）in（） 37返回（test_acc，round（l，5）） 38 ---＆GT; 39 define_batch（0.005） 40 run_batch（）
在define_batch（beta）中 11 shape =（batch_size，num_var）） 12 tf_train_labels = tf.placeholder（tf.float32，shape =（batch_size，num_labels）） ---＆GT; 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 常数（值，dtype，形状，名称） 159 tensor_value = attr_value_pb2.AttrValue（） 160 tensor_value.tensor.CopyFrom（ - ＆GT; 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 在make_tensor_proto中（values，dtype，shape） 320 nparray = np.array（values，dtype = np_dt） 321如果列表（nparray.shape）！= _GetDenseDimensions（values）： - ＆GT; 322引发ValueError（＆＃34;参数必须是密集张量：％s＆＃34;％值） 323＃python / numpy默认浮点类型是float64。我们更喜欢float32。 324 if（nparray.dtype == np.float64）且dtype为None：
ValueError：参数必须是密集张量：星期二 周三周四周五周六周日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 0.04608 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 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 0.0 0.0 227235 1.0 0.0 0.0 0.0 0.0 0.0 0.08787 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
以下是代码
的摘录# 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()
编辑： @gdhal感谢您的关注
train_dataset是一个pandas数据帧
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')
除了最后3个变量（Xnorm，Ynorm和Hournorm）之外，所有变量都是假人（取0或1值），这些变量是归一化为[0,1]区间的数值。 valid_dataset和test_dataset具有相同的格式
train_labels是一个熊猫系列
train_labels.describe()
count 790184
unique 39
top LARCENY/THEFT
freq 157434
Name: Category, dtype: object
valid_labels和test_labels具有相同的格式
答案 0 :(得分：3)
尝试输入numpy数组而不是pandas数据帧。