Hey guys, it's my first time doing a Zindi competition and I'm having trouble even breaking the all-zeros prediction benchmark score (~8.62 mae). Can somebody please give some tips and advice? Thank you
Use ensemble models, conduct eda, feature engineer and tune your model
use optuna for hyper paramter tuning. use cross validation. Remove unnecessary features. Use one-hot encoding for categorical variables.
Use ensemble models, conduct eda, feature engineer and tune your model
use optuna for hyper paramter tuning. use cross validation. Remove unnecessary features. Use one-hot encoding for categorical variables.