Criar modelos de ML com o BigQuery ML: laboratório com desafio avaliações

vivek k. · Revisado há about 1 year

Amanda B. · Revisado há about 1 year

Sijo Jose C. · Revisado há about 1 year

buggy

Joseph J. · Revisado há about 1 year

Susana D. · Revisado há about 1 year

Shivaji D. · Revisado há about 1 year

Mehdi M. · Revisado há about 1 year

Sivakumar V. · Revisado há about 1 year

krishna k. · Revisado há about 1 year

Done...

Miguel M. · Revisado há about 1 year

Krishna K. · Revisado há about 1 year

Krishna K. · Revisado há about 1 year

Travis H. · Revisado há about 1 year

Mehdi M. · Revisado há about 1 year

Alexander D. · Revisado há about 1 year

Andrew G. · Revisado há about 1 year

Unclear language. Also the instruction itself. Lack the details. I've lost several hours on trying to guess what is wrong and FAILED. Misleading parts of the instructions: - "Also you have to specify the model type as binary logistic regression... Use the model type as linear regression model" - "create a dataset with the dataset ID 'bq_dataset' ... Create a model ... in the precreated dataset 'bqml_dataset'"... with each attempt I was trying to create my model in each of these databases... It is not said which day partitions should be used for training when there are lots of them. And for evaluation. Evaluation by users, not explicitly said which column it is. userId and visitorId columns are always NULL. Google analytics is not known to every user, and it is not mentioned as a prerequisite for this exercise. That's why it's important to clearly describe the columns to be used. Or maybe the error messages could be more helpful. Now they say "create a model... in dataset"... and it's created but the check fails. Sometimes checking progress shows failure for some time, and after some time returns success without changing anything.

Wojciech C. · Revisado há about 1 year

Aldhitya P. · Revisado há about 1 year

task 4 always give me errors. Needs further instructions even though I have an output

Nikos L. · Revisado há about 1 year

Sometimes checking progress shows failure for some time, and after some time returns success without changing anything. You often ask to create a dataset e.g. 'bq_dataset' and then ask to create a model in different but similar dataset e.g. 'bqml_dataset'. It's misleading. Often it's not directly shown where the data is located (Google analytics tables) or not said which day partitions should be used for training when there are lots of them. This task assumes good knowledge of Google Analytics, but it's not introduced earlier... and it's not fair in my opinion.

Wojciech C. · Revisado há about 1 year

It was good but a bit repetived, i would like to create a model in a new dataset, maybe something simple like a iris datset or the housing dataset and leet the user chose his features

Antonio O. · Revisado há about 1 year

Lonnie Y. · Revisado há about 1 year

Takahide M. · Revisado há about 1 year

Divya S. · Revisado há about 1 year

Beatriz A. · Revisado há about 1 year