Research Article

An Empirical Study for Adopting Machine Learning Approaches for Gas Pipeline Flow Prediction

Table 10

The best performance of each method that we tried.

MeasureMeanLinear regressionSVRKNNNeural networkGBDT

MSE1.88 × 10832133.294.6 × 1051273.002426.29704.2363
MRE4.01 × 10−16.24 × 10−32.286 × 10−37.642 × 10−41.94 × 10−26.252 × 10−4
11.705 × 10−42.45 × 10−36.75 × 10−61.28 × 10−53.7 × 10−6