Realizar engenharia de atributos avançada no Keras avaliações

NICE

Sanal B. · Revisado há about 3 years

Shivam P. · Revisado há about 3 years

Yasin S. · Revisado há about 3 years

Carolina G. · Revisado há about 3 years

桢楠 蔡. · Revisado há about 3 years

Gabriel G. · Revisado há about 3 years

The transform function is a bit hard to comprehend. I suggest split it into two functions. The first one does the normalization and calculates euclidean distance. The second one creates the crossed feature.

Junhao Z. · Revisado há about 3 years

JAYAVARAPU SIVA T. · Revisado há about 3 years

Julia W. · Revisado há about 3 years

Wijaya P. · Revisado há about 3 years

good experience

saggurthi a. · Revisado há about 3 years

Some features, e.g., dayofweek and pickup_dropoff embedding features were defined but not used.

Sadeka I. · Revisado há about 3 years

Vishnu Teja R. · Revisado há about 3 years

Manjesh G. · Revisado há about 3 years

poornika b. · Revisado há about 3 years

Giles W. · Revisado há about 3 years

Marco C. · Revisado há about 3 years

very useful

sathish s. · Revisado há about 3 years

Ty S. · Revisado há about 3 years

Koji H. · Revisado há about 3 years

Xenon X. · Revisado há about 3 years

Toshi S. · Revisado há about 3 years

Just about enough time to complete. suggest to give 2 hours instead. Some questions: End of lab compares predicted taxi fare for both models with and without feature engineering. However there should be comparison back to either the actual taxi fare (if known) to see which model performs closer, or comparison using val RMSE for 2 models instead, or comparison of train/val loss/mse curves. Simply comparing 2 predicted taxi fares doesnt tell which performs better.

Joy L. · Revisado há about 3 years

Monjoy S. · Revisado há about 3 years

Jonathan C. · Revisado há about 3 years