Table 1: The 9 Measures Used to Quantify the Model Accuracy through Statistical Measurement.

No. Factor Formula Acceptable Indicators
1 Root Mean Square Error
(RMSE)
RMSE = 0 implies perfect fit.
2 Root Mean Square Percentage Error
(RMSPE%)
RMSPE% = 0 implies perfect fit.
Note: The percentage error cannot exceed 100%
3 Mean Absolute Deviation
(MAD)
MAD = 0 implies perfect fit.
4 Mean Bias Error
(MBE)
MBE < 0 = modelled count data is less than observed traffic volume, and
MBE > 0 = modelled count data is more than observed traffic volume.
5 Mean Percentage Error
(MPE%)
-10% < MPW < 10% = acceptability ranges.
6 Mean Absolute Percentage Error
(MAPE%)
MAPE% = 0 implies perfect fit.
Note: The percentage error cannot exceed 100%
7 Coefficient of Determination
(R2)
R2 is bounded, 0 ≤ R2 ≤ 1
R2 = 0 implies worst possible fit, and
R2 = 1 implies perfect fit.
8 The GEH Statistic
(GEH)
GEH < 5 = flows can be considered a good fit,
5 < GEH < 10 = flows may require further investigation, and
10 > GEH = flows cannot be considered to be a good fit.
9 Thiel’s U Statistic: URFORM
(Theil’s U)
U is bounded, 0 ≤ U ≤ 1
U = 0 implies perfect fit, and
U = 1 implies worst possible fit.