A novel approach to workload prediction using attention-based LSTM encoder-decoder network in cloud environment

EURASIP Journal on Wireless Communications and Networking

Table 5 The best-fitting history window length of different prediction sequence lengths; minimum well-fitting is the history window length that is the minimum in the well-fitting cases

Prediction sequence length	Best-fitting	Best-fitting RMSE×10⁻²	Minimum well-fitting
12	18	4.134	18
16	28	4.121	22
20	32	4.179	28
24	40	4.231	34