Research Article

Air Pollution Concentration Forecast Method Based on the Deep Ensemble Neural Network

Table 2

The accuracy comparisons between NN and NN+CLU.

ModelThe number of neurons of dense layersMAEMAPE

Group 1NN56.9919.59%
NN+CLU6.5617.77%
Group 2NN106.8919.67%
NN+CLU6.8719.65%
Group 3NN157.1319.89%
NN+CLU7.1219.78%
Group 4NN206.9519.19%
NN+CLU6.9219.11%
Group 5NN257.0019.96%
NN+CLU6.8319.00%
Group 6NN307.2120.29%
NN+CLU7.1520.14%