The article addresses the problem of methodological incorrectness in the comparative analysis of heterogeneous data, including data presented in aggregated form. Existing normalization methods are either inapplicable to aggregated data or fail to ensure interpretability and metric rigor.
A method is proposed–an extension of the author's adaptive multi-interval scale (AMIS)–as a software and methodological framework for normalizing and comparing aggregated data. Algorithms have been developed for converting aggregated data into representative samples (exact and optimized), along with an inverse transformation mechanism that establishes quantitative correspondences between original scales through the universal AMIS metric. The method was tested on three tasks: normalization and comparison of current academic grades, aggregated Unified State Exam results, and macroeconomic GDP indicators. The results demonstrate that AMIS creates a unified metric space for various data types, ensuring the correctness of arithmetic operations and statistically grounded correspondences between the original scales. The proposed approach solves the fundamental problem of integrating heterogeneous aggregated data. The open software suite (Python, C#, Excel) and verified data in repositories ensure full reproducibility of the results.