output_records.txt

~ 7000 data
Only one price:

LSTM MAE train: 0.024046414740739944
LSTM MAE val: 0.021323282090947356

After consider eight features:

LSTM MAE train: 0.020798744475696438
LSTM MAE val: 0.02001061281133617

LSTM MSQ train: 0.0012457879179005335
LSTM MSQ val: 0.0007795043391391522

Sailun's dataset:
LSTM MAE train: 2.8048360484955515
LSTM MAE val: 5.2174604515328795

LSTM MSQ train: 0.0012457879179005335
LSTM MSQ val: 0.0007795043391391522

After normalization:
LSTM MAE train: 0.018267576797983874
LSTM MAE val: 1.0227430657576329

LSTM MSQ train: 0.0006649258350362472
LSTM MSQ val: 2.127775191976839

After use L1 as criterion:
LSTM MAE train: 0.015229481473466104
LSTM MAE val: 1.0519369435960082

LSTM MSQ train: 0.0005034221462996769
LSTM MSQ val: 2.1006697224628317

~ 18000 data

LSTM MAE train: 0.04436826379937905
LSTM MAE val: 0.055107273727750795
LSTM MSQ train: 0.004540154328965787
LSTM MSQ val: 0.005959470631306754