Tensorflow data validation It turns out that there is a solut | L̶u̵m̶i̵n̷o̴u̶s̶m̶e̵n̵B̶l̵o̵g̵

Tensorflow data validation

It turns out that there is a solution from Google's TensorFlow Extended (TFX) platform for data validation. It can analyze training and service data to:

- compute and visualize descriptive statistics
- infer schema
- detect data anomalies

Within the TFX platform this works using the Apache Beam pipeline which is run using DirectRunner, the same pipeline can be propagated using other pipelines as well.

Check out Jupyter notebook and GCP's blueprint architecture for details

#big_data