Demand Forecasting for Automobile Spare Parts: A Comparative Study of ARIMA and Exponential Smoothing

Abstract

Effective inventory management is gaining increasing attention globally as businesses strive to optimize supply chain operations and reduce costs. In today’s competitive environment, where efficiency and sustainability are paramount, proper inventory management plays a critical role. It encompasses a wide range of processes, including inventory classification, prioritization, demand forecasting, and replenishment. This study integrates ABC-FSN analysis with advanced forecasting techniques—ARIMA (AutoRegressive Integrated Moving Average) and Exponential Smoothing—to enhance inventory control in medium-level automobile spare parts stores. ABC analysis categorizes inventory based on annual consumption value, while FSN analysis classifies items according to their movement rates (Fast-moving, Slow-moving, and Non-moving). ARIMA is employed to capture complex and irregular demand patterns, whereas Exponential Smoothing is utilized for stable and consistent demand trends. Using real-world data, the study compares the performance of both forecasting methods, demonstrating that ARIMA excels in handling complex demand patterns, while Exponential Smoothing is more effective for stable demand. The integration of classification and forecasting provides actionable insights for optimizing inventory levels, reducing stockouts, and minimizing holding costs. This framework not only enhances supply chain efficiency but also supports sustainable inventory management practices, contributing to a more resilient and future-ready supply chain.

Introduction

The study addresses inventory management challenges in medium-level automobile spare parts stores, which face variable demand, limited storage, and cost-service trade-offs. Efficient inventory management is crucial to ensure spare part availability, reduce holding costs, and prevent stockouts or overstocking.

The research focuses on combining inventory classification and demand forecasting techniques. ABC analysis categorizes parts by annual consumption value (high to low value), while FSN analysis classifies them by movement rates (fast, slow, non-moving). This dual classification aids prioritization and cost-effective inventory control.

For demand forecasting, the study applies two models: ARIMA (good for complex, irregular demand) and Exponential Smoothing (suitable for stable demand). Using real store data, the study compares these models’ forecasting accuracy to guide inventory decisions.

Key findings include:

• Exponential Smoothing outperforms ARIMA in accuracy and is better suited for medium-level stores with relatively stable demand patterns.

• Integrating ABC-FSN classification with forecasting helps optimize inventory by prioritizing high-value fast-moving parts and minimizing costs for low-value, slow/non-moving parts.

• This integrated approach improves operational efficiency, customer satisfaction, and cost management, making stores more resilient and sustainable.

The study also reviews relevant literature highlighting advances in machine learning and hybrid models for forecasting but notes gaps in focus on medium-level stores, integration of classification and forecasting, handling intermittent demand, real-world validation, and cost optimization.

The proposed methodology involves classifying inventory, forecasting demand using ARIMA and Exponential Smoothing, evaluating models by error metrics, and selecting the best approach per part.

Results show lower errors for Exponential Smoothing (MAE 2.42 vs 6.25, RMSE 2.81 vs 8.62), confirming its suitability for the target stores. The framework’s holistic management strategy enhances supply chain performance while controlling costs effectively.

Conclusion

This study proposes an integrated framework for inventory management and demand forecasting tailored to medium-level automobile spare parts stores. By combining ABC-FSN analysis with Exponential Smoothing and ARIMA, it optimizes inventory levels, reduces stockouts, and minimizes costs. Results indicate Exponential Smoothing excels for stable demand, while ARIMA suits complex patterns. This framework enhances inventory management and supports sustainability goals. Future research could explore external factors (e.g., economic conditions) and extend the framework to other industries for broader applicability.

References

[1] Vishwakarma et al. (2022). Predicting sales during COVID using Machine Learning Techniques. Journal of Supply Chain Management, 45(3), 123-135. [2] M. Rossi, A. et al. (2022). A Data-Driven Approach for Spare Parts Demand Forecasting in the Automotive Industry. International Journal of Production Economics, 210, 45-60. [3] El Filali et al. (2022). Machine Learning Applications in Supply Chain Management: A Deep Learning Model Using an Optimized LSTM Network for Demand Forecasting. Computers & Industrial Engineering, 165, 107-120. [4] K. Patel et al. (2022). Time Series Forecasting for Automotive Spare Parts Demand Using Prophet and LSTM. Journal of Forecasting, 41(2), 89-104. [5] Pacella et al. (2021). Evaluation of deep learning with long short-term memory networks for time series forecasting in supply chain management. Expert Systems with Applications, 180, 115-130. [6] R. Gupta et al. (2022). A Comparative Study of Machine Learning Models for Demand Forecasting of Automotive Spare Parts. International Journal of Logistics Management, 33(1), 45-60. [7] Bajaj et al. (2020). Sales Prediction Using Machine Learning Algorithms. Journal of Business Analytics, 12(4), 78-92. [8] S. Kumar (2021). Demand Forecasting for Automotive Spare Parts Using Machine Learning and Deep Learning Techniques. Journal of Operations Management, 39(2), 67-82. [9] J. Noh et al. (2020). Gated Recurrent Unit with Genetic Algorithm for Product Demand Forecasting in Supply Chain Management. Applied Soft Computing, 95, 106-120. [10] M. Ali et al. (2020). Intermittent Demand Forecasting for Automotive Spare Parts Using Croston’s Method and Its Variants. International Journal of Forecasting, 36(3), 45-60. [11] P. Wang (2021). Deep Learning for Intermittent Demand Forecasting in Automotive Aftermarket Supply Chains. Journal of Machine Learning Research, 22, 1-25. [12] Chopra, S., & Meindl, P. (2021).Supply Chain Management: Strategy, Planning,and Operation (8th Edition) Pearson Education.

Copyright

Copyright © 2025 Momin Abdullah , Dr. Apratul Chandra Shukla . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.