Performing Advanced Feature Engineering in BQML Reviews

Murilo A. · Reviewed almost 5 years ago

Chan S. · Reviewed almost 5 years ago

kazokada k. · Reviewed almost 5 years ago

Oskar F. · Reviewed almost 5 years ago

Iago A. · Reviewed almost 5 years ago

John K. · Reviewed almost 5 years ago

seunghwan o. · Reviewed almost 5 years ago

Eric T. · Reviewed almost 5 years ago

Works!

Alex C. · Reviewed almost 5 years ago

one slightly confusing aspect of the lab is that CREATE OR REPLACE MODEL is sensitive to the names of the columns provided in the query. As a result, when doing `CONCAT( dayoftheweek and hour as strings) as day_and_hour`, the RMSE will be different from the model trained with the same query, but with `CONCAT( dayoftheweek and hour as strings) as hourofday` this probably would go away with more data... but given the point where this appeared in the lab (swapping independent dayofweek and hour for the cross, expecting the model rmse to improve), it was kind of odd to see the rmse getting worse -- weirder when I figured WHY it was worse, and why it differed from the value previously present in the cell

Rodrigo K. · Reviewed about 5 years ago

Samuel Angel R. · Reviewed about 5 years ago

Lu Y. · Reviewed about 5 years ago

Kevin D. · Reviewed about 5 years ago

Christian C. · Reviewed about 5 years ago

Manu M. · Reviewed about 5 years ago

akhil s. · Reviewed about 5 years ago

Rahul D. · Reviewed about 5 years ago

Domenico S. · Reviewed about 5 years ago

Merin S. · Reviewed about 5 years ago

Xiaoya X. · Reviewed about 5 years ago

Mohamed G. · Reviewed about 5 years ago

Florian C. · Reviewed about 5 years ago

Kavan D. · Reviewed about 5 years ago

ROHITH K. · Reviewed about 5 years ago

Link L. · Reviewed about 5 years ago