Table 5 TFFN (SConv1D + earlyTGT) performance on the test set under different information sources
From: Traffic flow prediction based on depthwise separable convolution fusion network
FFN(SConv1D + earlyTGTR) | MAE | RMSE | R2(× 100) |
|---|---|---|---|
L | 188.1 | 254.2 | 42.8 |
L + W | 184.8 | 252.9 | 43.4 |
L + W + E | 168.9 | 242.9 | 47.8 |
L + W + E + T | 148.7 | 227.7 | 54.1 |