Innovation science and technologiy 1-jild 12-son (2025)
AI-BASED NORMALIZATION METHODOLOGY FOR COLLECTING AND PROCESSING KPI INDICATORS
Shuhratov, Mamurjon
Annotatsiya
The heterogeneity of employee performance data collected in organizations—stemming from variations informat, structure, and recording methods—creates significant inaccuracies within KPI systems. This article proposesan AI-based normalization methodology aimed at standardizing KPI data, automatically filtering noisy and inconsistententries, and converting heterogeneous inputs into a unified mathematical representation. The study employs NLPtechniques, min–max scaling, z-score standardization, Isolation Forest, and sentence-embedding models. Experimentalresults demonstrate that the proposed normalization pipeline increases data accuracy from 78% to 94% and reduces theKPI calculation time from 40 hours to 0.8 hours
normalization, artificial intelligence, data cleaning, automation, NLP.
Metadata manbasi: jurnal OAI-PMH arxivi · Sindex toʻliq matnni saqlamaydi, manbaga havola beradi.