Table 9 Average rank derived from the MCB test for DNN and statistical baseline models corresponding to the performance metrics (the best one is highlighted).

From: Salt rock creep deformation forecasting using deep neural networks and analytical models for subsurface energy storage applications

Datasets (MPa)

Models

Mean rank

RMSE

MAE

MAPE (%)

SMAPE (%)

5

N-BEATS

1

1

1

1

TCN

3

4

3

3

RNN

8

8

8

8

Transformer

4

3

4

4

ES

5

5

5

5

TBATS

6

6.5

6

6

Auto ARIMA

7

6.5

7

7

Theta

2

2

2

2

15

N-BEATS

1

1

1

1

TCN

3

3

3

3

RNN

8

7

8

7

Transformer

7

8

7

7

ES

6

6

6

6

TBATS

2

2

2

2

Auto ARIMA

4

4

4

4

Theta

5

5

5

5

25

N-BEATS

4

4

4

4

TCN

3

3

3

3

RNN

5

5

6

7

Transformer

8

7

7

6

ES

7

8

8

8

TBATS

3

2

2

2

Auto ARIMA

6

6

5

5

Theta

1

1

1

1

35

N-BEATS

3

3

2.5

2

TCN

2

2

2.5

3

RNN

8

6

8

6

Transformer

6

7

6

6

ES

1

1

1

1

TBATS

7

8

7

8

Auto ARIMA

4

4

4

4

Theta

5

5

5

5