7

I've followed the Tensorflow Reading Data guide to get my app's data in the form of TFRecords, and am using TFRecordReader in my input pipelines to read this data.

I'm now reading the guides on using skflow/tf.learn to build a simple regressor, but I can't see how to use my input data with these tools.

In the following code, the app fails on the regressor.fit(..) call, with ValueError: setting an array element with a sequence..

Error:

Traceback (most recent call last):
  File ".../tf.py", line 138, in <module>
    run()
  File ".../tf.py", line 86, in run
    regressor.fit(x, labels)
  File ".../site-packages/tensorflow/contrib/learn/python/learn/estimators/base.py", line 218, in fit
    self.batch_size)
  File ".../site-packages/tensorflow/contrib/learn/python/learn/io/data_feeder.py", line 99, in setup_train_data_feeder
    return data_feeder_cls(X, y, n_classes, batch_size)
  File ".../site-packages/tensorflow/contrib/learn/python/learn/io/data_feeder.py", line 191, in __init__
    self.X = check_array(X, dtype=x_dtype)
  File ".../site-packages/tensorflow/contrib/learn/python/learn/io/data_feeder.py", line 161, in check_array
    array = np.array(array, dtype=dtype, order=None, copy=False)

ValueError: setting an array element with a sequence.

Code:

import tensorflow as tf
import tensorflow.contrib.learn as learn

def inputs():
    with tf.name_scope('input'):
        filename_queue = tf.train.string_input_producer([filename])

        reader = tf.TFRecordReader()
        _, serialized_example = reader.read(filename_queue)

        features = tf.parse_single_example(serialized_example, feature_spec)
        labels = features.pop('actual')
        some_feature = features['some_feature']

        features_batch, labels_batch = tf.train.shuffle_batch(
            [some_feature, labels], batch_size=batch_size, capacity=capacity,
            min_after_dequeue=min_after_dequeue)

        return features_batch, labels_batch


def run():
    with tf.Graph().as_default():
        x, labels = inputs()

        # regressor = learn.TensorFlowDNNRegressor(hidden_units=[10, 20, 10])
        regressor = learn.TensorFlowLinearRegressor()

        regressor.fit(x, labels)
        ...

It looks like the check_array call is expecting a real array, not a tensor. Is there anything I can do to massage my data into the right shape?

Mark McDonald
  • 7,571
  • 6
  • 46
  • 53
  • What happens if you do x = x.eval() and labels = labels.eval() before your regressor.fit call? This should evaluate the tensor into an array, but I doubt this is the correct way of doing this with skflow... – mathetes May 30 '16 at 15:26
  • @mathetes, that seems to work but before I head down this path, is that the 'tf-y' way of doing things? My intuition is that the TF graph should be moving the data around, not my program. – Mark McDonald May 31 '16 at 09:37
  • Absolutely, I'm sorry I didn't specify but it was meant just as a way of debugging. That's why I posted a comment rather than an answer. I can't help you further though, I'm not familiar with skflow – mathetes May 31 '16 at 12:15

1 Answers1

1

It looks like the API that you were working with is depreciated. If you use a more modern tf.contrib.learn.LinearRegressor (I think >= 1.0), you are supposed to specify the input_fn, which basically produces the inputs and labels. I think in your example, that would be as simple as changing your run function to:

def run():
    with tf.Graph().as_default():
        regressor = tf.contrib.learn.LinearRegressor()
        regressor.fit(input_fn=my_input_fn)

and then defining an input function called my_input_fn. From the docs, this input function takes the form:

def my_input_fn():

    # Preprocess your data here...

    # ...then return 1) a mapping of feature columns to Tensors with
    # the corresponding feature data, and 2) a Tensor containing labels
    return feature_cols, labels

I think the documentation can get you the rest of the way. It is difficult from here for me to say how you should proceed without seeing your data.

Engineero
  • 12,340
  • 5
  • 53
  • 75