A big bug casuse two errors

[Borrk]

There is actually one fault that causes two errors:
the fault is when you load stock data, its order is ascending.
errors:

1. your training data is in reverse order, which might won't impact the result, but it is unreasonable in reality.
2. this error is actually series.
   when you merge the training and test data, the dates are not continuous as marked with the red circle in the figure.
   "dataset_total = pd.concat((dataset_train['Open'], dataset_test['Open']), axis = 0)"

actually, they should be re-ordered: dataset_test = dataset_test.sort_index(ascending=False).