Accurate characterization and structured interpretation of qubit errors are essential for advancing scalable quantum computation. In this work, we introduce a Category-Based Error Budgeting () framework for systematic analysis of qubit calibration data in noisy intermediate-scale quantum () processors. The proposed methodology decomposes the total error budget into physically and statistically meaningful categories, enabling quantitative evaluation of their relative contributions through two core metrics: the relative contribution rate ( ) and the disproportionality factor ( ).Beyond static categorization, the framework incorporates correlation-aware analysis by constructing covariance structures derived from two-qubit gate errors, allowing error interdependencies to be explicitly modeled. We further integrate the budgeting formalism into decoder weight assignment strategies and compare three decoding models: uniform weighting, individual error-based weighting, and a category-correlation-aware model. Logical error rates are estimated under each scheme to evaluate performance impact.The framework is implemented as an interactive graphical analysis tool and validated using real calibration datasets from contemporary superconducting quantum processors. Results demonstrate that category-level decomposition reveals structurally dominant error sources that are not evident from aggregate statistics alone, and that correlation-aware weighting provides improved logical error rate estimation.This work establishes a principled and operational bridge between calibration diagnostics and decoder optimization, offering a scalable methodology for structured error analysis in near-term quantum hardware.