Abstract:
Signalized intersection is one of the vital components of the entire road network. The
operational conditions of intersections considerably affect the performance of the whole road
network system. The performance level of signalized intersection is measured in terms of
Level of Service (LOS). Existing studies on LOS at signalized intersections are based on
conventional linear regression (CLR) techniques and those models failed to estimate accurate
LOS of signalized intersections due to basic assumptions of CLR methods. This paper
explores the fundamentals of most popular fuzzy linear regression (FLR) techniques and the
application of FLR methods in developing the LOS model at signalized intersections. The
proposed methodology derived in two steps. First step, membership function developed and
the fuzzy input values defuzzified in crisp value by applying the centroid method. Second
step, the fuzzy least square method is applied to develop the required model. The proposed
methodology is applied in Pedestrian LOS model development. Finally, mean absolute
percentage error values are compared between conventional regression and fuzzy regression
models and the results shown that fuzzy regression models provide more precise and reliable
solutions. The proposed new methodology can be used to develop a LOS model in the
transportation field.
