Beyond conventional performance assessment of recycled aggregate concrete through integrated data analytics

The mechanical performance of recycled aggregate concrete (RAC) varies with aggregate quality, mixture composition, and experimental conditions, complicating the interpretation of published replacement-ratio trends. This study integrates descriptive statistics, LOWESS smoothing, bootstrap uncertainty estimation, derivative-based analysis, and clustering to examine five normalized indicators: compressive strength, splitting tensile strength, flexural strength, bulk density, and elastic modulus. Property-specific complete-case analyses cover RCA replacement ratios from 0 to 100%. The fitted relationships indicate overall reductions relative to the reference mixtures, together with property-specific non-monotonic features. Elastic modulus exhibits the largest between-category effect-size estimate, whereas bulk density shows comparatively small absolute changes in normalized response. The maximum-absolute-slope locations differ among properties and do not consistently represent deterioration: the NCS location occurs on a local upward segment of the bootstrap-mean curve. These locations and the associated high-slope intervals are therefore interpreted as method-dependent descriptors rather than physical failure thresholds. Four fixed RCA intervals provide an exploratory description of the compiled evidence, without defining safe replacement limits or structural application classes. The framework supports transparent synthesis and hypothesis generation, while source heterogeneity, within-study dependence, uneven coverage, and sensitivity to the analytical specification limit broader inference.