temporian.to_tensorflow_dataset #
to_tensorflow_dataset(
evset: EventSet, timestamps: str = "timestamp"
) -> tensorflow.data.Dataset
Converts an EventSet
to a tensorflow Dataset.
Usage example
evset = event_set(
timestamps=[1, 2, 3, 4],
features={
"f1": [10, 11, 12, 13],
"f2": [b"a", b"b", b"c", b"d"],
"label": [0, 1, 0, 1],
},
)
tf_dataset = tp.to_tensorflow_dataset(evset)
def extract_label(example):
label = example.pop("label")
return example, label
tf_dataset = tf_dataset.map(extract_label).batch(100)
model = ... # A Keras model
model.fit(tf_dataset)
Parameters:
Name | Type | Description | Default |
---|---|---|---|
evset |
EventSet
|
Input event set. |
required |
timestamps |
str
|
Output key containing the timestamps. |
'timestamp'
|
Returns:
Type | Description |
---|---|
Dataset
|
TensorFlow dataset created from EventSet. |