Using BigQuery ML to Predict Penguin Weight Reviews

Same data is used for training, evaluation, and prediction. This is a terrible anti-pattern. Plus, because there are fewer than 500 records, BigQuery uses the whole dataset for training, the default splitting method does not work in this scenario. The entire dataset is consumed for training. I would advise to update the dataset, the commands, or the lab instructions to make it clear what's going on with the points that I had just mentioned. Also might be worth mentioning that for datasets containing fewer than 10,000 examples bigquery linear regression model uses analytical solution, not gradient descent. I believe this is documented.

Oliver F. · Reviewed over 2 years ago

Ahmad K. · Reviewed over 2 years ago

good

Yuliana A. · Reviewed over 2 years ago

Lary K. · Reviewed over 2 years ago

Bryam A. · Reviewed over 2 years ago

ok

Hugo A. · Reviewed over 2 years ago

Hugo A. · Reviewed over 2 years ago

Bogdan M. · Reviewed over 2 years ago

MacKenzie C. · Reviewed over 2 years ago

Pavan V. · Reviewed over 2 years ago

Sam L. · Reviewed over 2 years ago

Fozia Z. · Reviewed over 2 years ago

Lee C. · Reviewed over 2 years ago

Sadegh T. · Reviewed over 2 years ago

Michael C. · Reviewed over 2 years ago

Nurmukhamed U. · Reviewed over 2 years ago

Seamus J. · Reviewed over 2 years ago

Caio F. · Reviewed over 2 years ago

Pritam K. · Reviewed over 2 years ago

Sophie B. · Reviewed over 2 years ago

Swapnadip P. · Reviewed over 2 years ago

Rodrigo L. · Reviewed over 2 years ago

Kuldeep v. · Reviewed over 2 years ago

Ana Rita F. · Reviewed over 2 years ago

Manuel M. · Reviewed over 2 years ago