tf.feature_column.categorical_column_with_hash_bucket

View source on GitHub

Represents sparse feature where ids are set by hashing.

tf.feature_column.categorical_column_with_hash_bucket(
    key,
    hash_bucket_size,
    dtype=tf.dtypes.string
)

Use this when your sparse features are in string or integer format, and you want to distribute your inputs into a finite number of buckets by hashing. output_id = Hash(input_feature_string) % bucket_size for string type input. For int type input, the value is converted to its string representation first and then hashed by the same formula.

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.

Example:

import tensorflow as tf
keywords = tf.feature_column.categorical_column_with_hash_bucket("keywords",
10000)
columns = [keywords]
features = {'keywords': tf.constant([['Tensorflow', 'Keras', 'RNN', 'LSTM',
'CNN'], ['LSTM', 'CNN', 'Tensorflow', 'Keras', 'RNN'], ['CNN', 'Tensorflow',
'LSTM', 'Keras', 'RNN']])}
linear_prediction, _, _ = tf.compat.v1.feature_column.linear_model(features,
columns)

# or
import tensorflow as tf
keywords = tf.feature_column.categorical_column_with_hash_bucket("keywords",
10000)
keywords_embedded = tf.feature_column.embedding_column(keywords, 16)
columns = [keywords_embedded]
features = {'keywords': tf.constant([['Tensorflow', 'Keras', 'RNN', 'LSTM',
'CNN'], ['LSTM', 'CNN', 'Tensorflow', 'Keras', 'RNN'], ['CNN', 'Tensorflow',
'LSTM', 'Keras', 'RNN']])}
input_layer = tf.keras.layers.DenseFeatures(columns)
dense_tensor = input_layer(features)