Scalable Analytics for Enterprise Decisions: From MapReduce to Holiday-Aware Demand Forecasting

Volume 7 of 10 in the Engineering-to-Research Monograph Series. Enterprises do not lack data; they lack the path from data to decisions. That path has two halves usually taught and tooled apart: the infrastructure half (MapReduce, Spark, distributed storage) and the modeling half (turning features into accurate, trustworthy predictions). This report joins them into a single decision-oriented analytics workflow and argues a deliberately unfashionable thesis: for the tabular data on which most enterprise decisions rest, scale should be spent on feature quality and validation, not model complexity, and tree-ensemble models remain the rational default. The argument is grounded in two implementation studies: a holiday-aware retail demand-forecasting pipeline with engineered calendar and macroeconomic features, and an Extremely Randomized Trees model for vehicle-price prediction emphasizing feature importance for decision support. It contributes a five-stage data-to-decision workflow, a feature-first design principle, and an evaluation of modeling approaches for tabular enterprise data.The paper and figures are licensed CC BY 4.0; companion code is released under the MIT License. This work contains no confidential or proprietary employer information.