DESIGNING A MARKET DECISION STACK FOR FAIR BUDGET ALLOCATION IN INDIA FMCG

Abstract

The FMCG national budget for India must be allocated to state-category cells which are not homogeneous. Affordability, digital preparedness, CPI pressure, festival dates, and advertisement clutter differ. Suitability charts provide us with regions that present opportunity and money flows there. In this study, we end this process with a constrained ₹ allocation an examiner can audit, and with a minimum allocation threshold that will not exclude smaller markets from consideration. Public data sources come in as a harmonized secondary panel by state × category × month. Opportunity Index (OI) measures structural suitability. Growth propensity captures the short-term dynamics. Finally, logistic regression, Random Forest, XGBoost, and SVM algorithms compete in a time cross-validation. ROC-AUC and average precision select a winner among them. The combined scores determine Investment/Protect/Probe/Suppress decision. An illustrative budget is sliced into EV share values. Allocation fairness thresholds and digital caps follow. One experiment writes the playbook and the thesis. Panel data goes to Data & Appendix-I. The linear model defeated the tree learners in the bake-off for holdout. In instances where OI and propensity were congruent, the Invest cells consumed a big chunk of the sample budget. The floors pushed expenditure towards more restricted eligible markets. The EV ranking order remained the same. The counts and rupee differences are in Results. The board receives an audit trail from total public numbers to priority positions and restricted Rupee allocations, cell by cell. Up next is to anchor portal files to actual releases and sharper category testing.