2

I am trying to visualize model results using tfma.view.render_slicing_metrics. I am getting the results with tfma.load_eval_result(tfma_artifact.uri) and I am able to print the results. 

[EvalResult(slicing_metrics=[((('trip_start_hour', 2),), {'precision': {'doubleValue': 0.6666666865348816}, 'accuracy_baseline': {'doubleValue': 0.8188405632972717},
...................................
'predictioexample_weight_metric_key='post_export_metrics/example_count', num_bootstrap_samples=1))]

But when I try to visualize the results with tfma.view.render_slicing_metrics(tfma.load_eval_result(tfma_artifact.uri)) I am seeing a empty visualization.

import tensorflow_model_analysis as tfma
tfma_artifact = store.get_dest_artifact_of_type(11,'ModelEvalPath')
tfma.load_eval_result(tfma_artifact.uri)
tfma.view.render_slicing_metrics(tfma.load_eval_result(tfma_artifact.uri))
giser_yugang
  • 6,058
  • 4
  • 21
  • 44
Rathna Subramanian.T.A
  • 21
  • 2
  • Check your browser logs. That might be an issue loading the Jupyter extension. – Benjamin Tan Wei Hao Jun 28 '19 at 09:19
  • 1
    For me this was an issue with running TensorFlow, TensorBoard and JupyterNotebook in a docker container. Ports 6006, 6007, 6008 ... etc. get used in ascending order and these need to be exposed to allow the browser to connect. – Reg Whitton Jan 08 '20 at 15:59

0 Answers0