Table 6 Diebold–Mariano test results of the reference model and the proposed model on the different datasets.

From: Novel hybrid data-driven modeling based on feature space reconstruction and multihead self-attention gated recurrent unit: applied to PM2.5 concentrations prediction

Model

Lanzhou

Xi’an

Beijing

Shijiazhuang

Chengdu

Lhasa

CNN

3.7492*

112.7258*

8.2718*

139.4179*

124.5073*

134.2444*

Elman

3.2273**

113.3857*

85.5030*

117.2315*

134.7368*

114.5253*

LSTM

3.4162*

113.5411*

84.9819*

129.8716*

130.1819*

109.3463*

BiLSTM35

3.1751**

119.2811*

81.3544*

92.6590*

134.8324*

125.0313*

GRU

3.2959*

104.1842*

83.5771*

142.4837*

130.0248*

106.4364*

MSAGRU

3.2685**

99.8772*

86.9596*

145.8403*

138.6332*

104.0910*

CNN-GRU 56

3.6621*

102.6903*

7.1443*

9.8557*

120.9061*

129.3515*

CEEMDAN-GRU

73.6663*

118.2614*

89.3533*

148.6870*

110.4164*

76.3547*

STL-MSAGRU

74.0572*

92.9375*

77.3538*

126.1690*

6.7944*

96.1930*

STL-LSTM

51.8039*

94.8233*

4.5786*

118.6124*

8.2040*

94.8237*

STL2-LSTM

61.3682*

99.6880*

4.8423*

135.4397*

9.2011*

91.0683*

3D CNN-GRU36

4.1259*

92.2599*

120.6974*

139.8668*

128.5119*

99.4922*

EEMD-ALSTM69

79.9051*

107.8205*

77.1872*

138.4936*

10.4169*

74.4446*

FSR-AGRU

–

–

–

–

–

–

  1. “ *” indicates 0.01 significance level, “ **” indicates 0.05 significance level.