关于“通过 BigQuery ML 创建机器学习模型：实验室挑战赛”的评价

Paul Cyrille H. · 已于 9 months前审核

Aaron K. · 已于 9 months前审核

Pankaj B. · 已于 9 months前审核

JOSE E. · 已于 9 months前审核

JOSE E. · 已于 9 months前审核

great lab

Karim A. · 已于 9 months前审核

TANIA ALEJANDRA L. · 已于 9 months前审核

JOSE E. · 已于 9 months前审核

Mohammad R. · 已于 9 months前审核

JOSE E. · 已于 9 months前审核

Very disappointing. 1) The dataset is not clearly specified, the features neither (is the goal to check per unique visitor or per session?) 2) in the assignment, two contradicting instructions are given: "Also you have to specify the model type as binary logistic regression" and 2 lines below "Use the model type as linear regression model" - the second doesn't make sense for binary classification (will or won't buy). 3) No specification on how to deal with train-validation split; some labs have the split, some don't, it's ambiguous 4) even after trying all combinations (even with the linear regression, i.e. fitting the number of transactions as if it were float) the result gets NOT verified by the module. I got the data and trained a model myself separated from GCP and got THE SAME results as here, so obviously the model is well trained. Sad work out there, sorry.

Darina L. · 已于 9 months前审核

João T. · 已于 9 months前审核

krishna b. · 已于 9 months前审核

Pushpendra Y. · 已于 9 months前审核

Takahide M. · 已于 9 months前审核

Rajeswari P. · 已于 9 months前审核

Girish K. · 已于 9 months前审核

Denise T. · 已于 9 months前审核

Aditya S. · 已于 9 months前审核

Great

Pramod K. · 已于 9 months前审核

Asif M. · 已于 9 months前审核

It is important to make the instruction clearer stating that the student is to use already trained models and not to train another subscriber_type model. It can be confusing

GALINA K. · 已于 9 months前审核

Fredrick Barasa O. · 已于 9 months前审核

Changming C. · 已于 9 months前审核

Smita W. · 已于 9 months前审核