apache/beam

[Task]: Create a script to train sklearn model for IT test.

Open

Aperta il 5 gen 2023

Vedi su GitHub
 (19 commenti) (0 reazioni) (4 assegnatari)Java (7313 star) (4097 fork)batch import
P2good first issuepythonrun-inferencetask

Descrizione

What needs to happen?

Sklearn doesn't offer backward compatibility wrt models on newer versions. In the Sklearn IT tests, we use models trained manually and this can get outdated when there is an update to sklearn.

So to tackle this, we need to create a script which trains the sklearn models on the data and then publish this model to a GCS bucket, once this is done we can use this model to run Sklearn IT test.

Issue Priority

Priority: 2 (default / most normal work should be filed as P2)

Issue Components

  • Component: Python SDK
  • Component: Java SDK
  • Component: Go SDK
  • Component: Typescript SDK
  • Component: IO connector
  • Component: Beam examples
  • Component: Beam playground
  • Component: Beam katas
  • Component: Website
  • Component: Spark Runner
  • Component: Flink Runner
  • Component: Samza Runner
  • Component: Twister2 Runner
  • Component: Hazelcast Jet Runner
  • Component: Google Cloud Dataflow Runner

Guida contributor

Tech stack
pythongcp
Dominio
testingdevops
Tipo issue
test
DifficoltàQuanto dovrebbe essere impegnativa per un nuovo contributor.
3
Tempo stimatoTempo stimato per completare e verificare uno scope piccolo.
1-2 days
Stato attivitàQuanto è probabile che la issue sia ancora attiva e reviewable.
active
ChiarezzaQuanto chiaramente la issue descrive problema, scope e risultato atteso.
clear
Prerequisiti
PythonsklearnGCSBeam testing infra
Adatta ai principiantiQuanto la issue è adatta a contributor alla prima esperienza.
60
Direzione di ricerca
Examine existing IT tests for sklearn models (likely in the Python SDK test directory). Check if there is a GCS bucket already used for test artifacts. The script should load training data, train a model with sklearn, and upload to a designated GCS location. Review comments on issue #24903 for any discussions on implementation details.