Research Article

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

Table 1

Sensitivity tests using different regularization methods and penalty strength.

Alpha1 × 10−61 × 10−50.00010.0010.010.11.0

LassoMSE4 × 1044 × 1044.2 × 1046 × 1041.1 × 1052.9 × 1061.9 × 108
MRE7.2 × 10−37.2 × 10−37.4 × 10−39.8 × 10−31.3 × 10−25.3 × 10−24.0 × 10−1
2.1 × 10−42.1 × 10−42.2 × 10−43.1 × 10−45.7 × 10−41.5 × 10−29.9 × 10−1

RidgeMSE3.2 × 1043.2 × 1043.2 × 1043.2 × 1043.2 × 1043.2 × 1043.2 × 104
MRE6.2 × 10−36.2 × 10−36.2 × 10−36.2 × 10−36.2 × 10−36.2 × 10−36.2 × 10−3
1.7 × 10−41.7 × 10−41.7 × 10−41.7 × 10−41.7 × 10−41.7 × 10−41.7 × 10−4