Measurement-Type "Calibration" of Expert Estimates Improves Their Accuracy and Their Usability: Pavement Engineering Case Study

Authors

Edgar Daniel Rodriguez Velasquez, The University of Texas at El PasoFollow
Carlos M. Chang Albitres, The University of Texas at El PasoFollow
Vladik Kreinovich, The University of Texas at El PasoFollow

Publication Date

9-2018

Comments

Technical Report: UTEP-CS-18-51a

To appear in Proceedings of the IEEE Symposium on Computational Intelligence for Engineering Solutions CIES'2018, Bengaluru, India, November 18-21, 2018.

Abstract

In many applications areas, including pavement engineering, experts are used to estimate the values of the corresponding quantities. Expert estimates are often imprecise. As a result, it is difficult to find experts whose estimates will be sufficiently accurate, and for the selected experts, the accuracy is often barely within the desired accuracy. A similar situations sometimes happens with measuring instruments, but usually, if a measuring instrument stops being accurate, we do not dismiss it right away, we first try to re-calibrate it -- and this re-calibration often makes it more accurate. We propose to do the same for experts -- calibrate their estimates. On the example of pavement engineering, we show that this calibration enables us to select more qualified experts, and make estimates of the current experts more accurate.