Enhancing Stock Price Trend Prediction via a Time-Sensitive Data Augmentation Method

Table 2

Comparison experiments by adding random noises with diverse scales over both the high-frequency subseries and original time series.

	Random noise		Random noise
	High frequency (ours)		Original time series
Scale coefficient ()	MSE	MAE	MSE	MAE
0.05	5.147	0.934	5.142	0.939
0.1	5.111	0.934	5.418	0.986
0.15	5.205	0.931	5.313	0.942
No data augmentation			5.106	0.923