Laserfiche WebLink
%In shows the percent of assigned volumes within allowable errors <br />specifications. <br />R2, the coefficient of determination or 'goodness of fit' statistic, shows how well <br />the regression line represents the assignment data. <br />There are no national standards for R2 or RMSE. However, there are guidelines <br />that have been established by Caltrans for data used in air quality analysis. The <br />guidelines recommend an R2 of 0.88, a maximum RMSE of 35%, and a minimum <br />%In of 75% for links classified as Principal Arterials and above. <br />Analysis of all links shows an R2 of 0.92 and of Arterials and above shows an R2 <br />of 0.89, which are both better than the guideline of 0.88. Typically we see higher <br />values of R2 when there is a broader range of counts. If analyzing just higher <br />volume facilities, the counts are more clustered and thus it is more difficult to <br />plot these in a straight line. <br />The model also shows a %RMSE of 41 % for all links and a %RMSE of only 23% <br />for Arterials and above. The guideline of 35% or less is only applicable for <br />Principal Arterials and above, so the 23% looks very good. <br />Analyzing all link classes shows the 84% of the count locations within acceptable <br />bounds. The analysis of Arterials and above shows 93% of the count locations are <br />within bounds. This is much better than the recommended standard of 75% for <br />only links classified as Principal Arterial or above. <br />Slope of the line in both cases is very close to 1.0, with values ranging from 0.95 <br />to 1.04. This is considered to be very good. <br />The 2008 Kittitas County model is well calibrated and can be used for forecasts. <br />Kittitas County, Washington Transportation Model -May 2009 Page 44