New Paper Maps Excel-to-Pandas Migration for Reproducible Data Workflows

A new academic paper by Sidney Shapiro, Daniel Pearson, and Emiliano Sebastian Gonzalez Venegas, posted on arXiv (2606.14924), tackles a persistent pain point in business analytics: the fragility of spreadsheet-heavy workflows. The authors argue that while Excel remains a dominant tool for ad-hoc analysis, workbooks grown through formulas, manual edits, and copy-paste refreshes become nearly impossible to audit, reproduce, or govern at scale. The paper positions the Python pandas library as a practical transformation layer—not a wholesale replacement for Excel—that preserves familiar table concepts while making every assumption explicit. It treats pandas as a bridge that lets analysts keep Excel as a friendly input/output interface for stakeholders while moving the core analysis into code that is repeatable, version-controlled, and defensible.

The paper contributes several concrete resources: a full Excel-to-pandas migration mapping (e.g., VLOOKUP → merge, pivot tables → groupby), a taxonomy of nine common workflow categories (from simple filtering to multi-source joins), seven end-to-end examples drawn from real business analytics and applied research, and a failure-mode catalog that warns of pitfalls like misaligned indices, silent type coercion, and memory blow-up. The authors also include reusable code recipes for governed tabular work. With 39 pages and 8 figures, this paper is designed to be a practical reference for analysts transitioning from spreadsheets to programmable, reproducible analysis—without requiring them to abandon Excel entirely. For any team struggling to audit or scale their spreadsheet processes, this guide offers a concrete, battle-tested path forward.