Analogy-based entropic similarity between observations and/or predictions (AESOP)
Salvador Eugenio C. Caoili · 2015
Benchmarking the accuracy of predictive tools enables their progressive development. For this purpose, various measures of accuracy (e.g., the Pearson correlation coefficient [PCC] and the area under the receiver operator characteristic curve [AUROCC]) are employed, albeit typically for use with data representing either continuous or dichotomous variables only (as exemplified by PCC and AUROCC, respectively); yet data representing both of these variable types may be analyzed using an alternative measure, namely analogy-based entropic similarity between observations and/or predictions (AESOP).