/TensorFlow 2.4


A CategoricalColumn that returns identity values.

Use this when your inputs are integers in the range [0, num_buckets), and you want to use the input value itself as the categorical ID. Values outside this range will result in default_value if specified, otherwise it will fail.

Typically, this is used for contiguous ranges of integer indexes, but it doesn't have to be. This might be inefficient, however, if many of IDs are unused. Consider categorical_column_with_hash_bucket in that case.

For input dictionary features, features[key] is either Tensor or SparseTensor. If Tensor, missing values can be represented by -1 for int and '' for string, which will be dropped by this feature column.

In the following examples, each input in the range [0, 1000000) is assigned the same value. All other inputs are assigned default_value 0. Note that a literal 0 in inputs will result in the same default ID.

Linear model:

video_id = categorical_column_with_identity(
    key='video_id', num_buckets=1000000, default_value=0)
columns = [video_id, ...]
features = tf.io.parse_example(..., features=make_parse_example_spec(columns))
linear_prediction, _, _ = linear_model(features, columns)

Embedding for a DNN model:

columns = [embedding_column(video_id, 9),...]
features = tf.io.parse_example(..., features=make_parse_example_spec(columns))
dense_tensor = input_layer(features, columns)
key A unique string identifying the input feature. It is used as the column name and the dictionary key for feature parsing configs, feature Tensor objects, and feature columns.
num_buckets Range of inputs and outputs is [0, num_buckets).
default_value If set, values outside of range [0, num_buckets) will be replaced with this value. If not set, values >= num_buckets will cause a failure while values < 0 will be dropped.
A CategoricalColumn that returns identity values.
ValueError if num_buckets is less than one.
ValueError if default_value is not in range [0, num_buckets).

© 2020 The TensorFlow Authors. All rights reserved.
Licensed under the Creative Commons Attribution License 3.0.
Code samples licensed under the Apache 2.0 License.