Prediction of Bridge Component Ratings Using Ordinal Logistic Regression Model

Table 6

Model comparison statistics with 2013 data.

Model Statistics	Multiple Linear Regression			Ordinal Regression
	Deck	Superstructure	Substructure	Deck	Superstructure	Substructure
Sample Size	2309	2943	2943	2309	2943	2943
Sum of Absolute Residuals	1568	1832	2316	1360	1480	2207
Sum of Residual Squares	2328	2642	4072	1830	1950	3677
Bias	1.096	1.099	1.147	1.023	1.022	1.097
Accuracy	1.126	1.119	1.202	1.099	1.081	1.2
Out-of-range Forecasts	0%	0.2%	0.2%	0%	0%	0%
Percent Exact Forecasts	45%	48%	42%	49%	56%	43 %
Percent Estimation within 1 Rating Difference	88%	90%	84%	93%	94%	87%
Percent Estimation within 2 Rating Differences	98%	97%	94%	100%	100%	97%