Citation:

Vikram Tihal, Dr. Jagat Naryan Giri. "Ensemble Demand Forecasting for Perishable Inventory in Quick-Commerce: A Comparative Analysis with Weather and Behavioral Variables" International Research Journal of Economics and Management Studies, Vol. 5, No. 1, pp. 109-119, 2026. Crossref. http://doi.org/10.56472/25835238/IRJEMS-V5I2P114

Abstract:

Predicting demand for perishable goods is the most critical financial challenge for India’s quick-commerce sector, where margins are thin, and delivery windows are tight. This study addresses the "spoilage vs. stockout" dilemma by analyzing a massive dataset of 1.2 million daily orders spanning 21 months across 180 dark stores in Delhi-NCR, Mumbai, and Bengaluru. By comparing traditional statistical methods against modern machine learning and deep learning, the research shows a novel SARIMA-XGBoost hybrid model as the best solution. This hybrid approach achieves a Mean Absolute Percentage Error (MAPE) of 14.1%, significantly outperforming traditional Exponential Smoothing (24.8%) and SARIMA (19.3%) baselines. Unlike complex LSTM models that require heavy computation, this hybrid model generates predictions in just 85 milliseconds, making it fast enough for real-time inventory decisions. The analysis also confirms that while competitor pricing has a negligible impact on immediate demand, local weather variables like temperature and rainfall are critical drivers for fresh produce sales. Implementation of this weather-aware model in a simulated 100-store network reduced stockouts from 3.8% to 1.2% and lowered spoilage by over 7 percentage points. These operational efficiencies translate to a total annual financial benefit of ₹47.11 crore, offering a sustainable path to profitability for dark store operators.

References:

[1] Al Mesfer, A. S. (2023). Forecast-driven inventory management for the fast-moving consumer goods (FMCG) industry. MIT D-Space Thesis, Master of Science in Supply Chain Management.
[2] Box, G. E., & Jenkins, G. M. (1970). Time series analysis: Forecasting and control. Holden-Day.
[3] Brown, R. G. (1963). Smoothing, forecasting and prediction of discrete time series. Prentice-Hall.
[4] Chen, F., Drezner, Z., Ryan, J. K., & Simchi-Levi, D. (2000). Quantifying the bullwhip effect in a supply chain. Journal of Operations Management, 17(3), 309-323.
[5] Falatouri, T., Kuhn, J., Ranz, S., Schwab, M., & Strohm, J. (2022). A data science approach for demand forecasting in the FMCG supply chain. In Proceedings of 2022 the 8th International Conference on Supply Chain Management (pp. 18-27).
[6] Ganapathy, V., & Gupta, C. (2023). Critical success factors for quick commerce grocery delivery in India. In Exploring New Frontiers in Commerce and Management (Vol. 1, Issue 1, pp. 80–85). SGOC USI Publication.
[7] Holt, C. C. (1960). Forecasting seasonals and trends by exponentially weighted moving averages. International Journal of Forecasting, 20(1), 5-10.
[8] Liu, X., Terai, H., & Ren, F. (2019). Impact of weather on consumer food purchases in Japan. Food Quality and Preference, 73, 95-103.
[9] Makridakis, S., Spiliotis, E., & Assimakopoulos, V. (2022). Statistical and machine learning forecasting methods: Concerns and ways forward. PLOS ONE, 17(3), e0194889.
[10] Nestorov, S., & Chowdhury, S. (2020). Temporal dynamics of consumer demand during festivals: Evidence from Indian e-commerce. International Journal of Consumer Studies, 44(6), 572-582.
[11] Petropoulos, F., Apiletti, D., Assimakopoulos, V., et al. (2020). Forecasting: Theory and practice. International Journal of Forecasting, 36(3), 640-688.
[12] Raj, G., Dada, M., & Roy, D. (2024). Competitive dynamics in quick commerce in India: A game-theoretic approach. IIM Ahmedabad Research.
[13] Ranjekar, G., & Roy, D. (2023). Rise of quick commerce in India: Business models and infrastructure requirements. IIMA Centre for Transportation and Logistics.
[14] Silver, E. A., Pyke, D. F., & Peterson, R. (1998). Inventory management and production planning and scheduling (3rd ed.). Wiley.
[15] Sivarajah, U., Irani, Z., Weerakkody, V., & Kapoor, K. K. (2019). Critical success factors of E-commerce platforms for SMEs in the context of the digital economy. Journal of Global Information Management, 27(2), 87-116.
[16] Tanvir, M. D., et al. (2025). Demand forecasting in retail using machine learning and big data. International Journal of Advanced Research in Computer and Communication Engineering, 14(3).
[17] Thijs, R., Pals, H., & Wouter, S. (2017). Weather sensitivity of grocery purchases: Implications for food demand forecasting. European Journal of Operational Research, 263(2), 565-577.

Keywords:

Demand Forecasting, Perishable inventory, Quick Commerce, SARIMA-XGBoost, Weather Covariates.