TensorFlow Transform

E457351

TensorFlow Transform is a TensorFlow-based library for performing scalable, full-pass data preprocessing and feature engineering that can be applied consistently in both training and serving.

Try in SPARQL Jump to: Surface forms Statements Referenced by

All labels observed (1)

Label Occurrences
TensorFlow Transform canonical 1

Statements (47)

Predicate Object
instanceOf data preprocessing library
feature engineering library
machine learning library
software library
basedOn TensorFlow NERFINISHED
compatibleWith TensorFlow 2.x NERFINISHED
TensorFlow Extended 1.x and later
developedBy Google NERFINISHED
TensorFlow team NERFINISHED
documentation https://www.tensorflow.org/tfx/transform
domain data engineering
machine learning infrastructure
ensures training-serving skew reduction
feature analyzes full training dataset
applies transformations at training time
computes data statistics
exports transformations for serving
integrates with TensorFlow models
supports Apache Beam runners
supports TensorFlow SavedModel export of preprocessing
supports batch data processing
supports bucketization
supports categorical feature handling
supports feature scaling
supports large-scale data processing
supports missing value handling
supports normalization
supports schema-based transformations
supports vocabulary computation
license Apache License 2.0
maintainer TensorFlow Extended team NERFINISHED
partOf TensorFlow Extended NERFINISHED
programmingLanguage Python
purpose consistent preprocessing for training and serving
feature engineering for machine learning
full-pass data preprocessing
scalable data preprocessing
repository https://github.com/tensorflow/transform
runsOn cloud data processing platforms via Beam runners
distributed data processing backends
shortName TFX Transform NERFINISHED
tf.Transform NERFINISHED
supports Apache Beam NERFINISHED
TensorFlow Extended pipelines NERFINISHED
usedFor offline training data preparation
online serving data preprocessing
production ML pipelines

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

TensorFlow Extended usesLibrary TensorFlow Transform