kaggle Ion-Switcing

private rank : 79
This code was created from Mobassir's public kernel base.

model : Wavenet

optimizer : AdamW

FE(Feature Engineering)

power signal, bi-directional shift signal

# power signal
power_signal = signal**2

# bi-directional shift signal per batch
df['signal_shift_pos_' + str(window)] = df.groupby('group')['signal'].shift(window).fillna(0)
df['signal_shift_neg_' + str(window)] = df.groupby('group')['signal'].shift(-1 * window).fillna(0)

Ensemble : Flip model ensemble

1. Forward model forward X / forward y
2. Reverse model reverse X / reverse y

The output of the two models was averaged and submit.