運用基於機器學習演算法之混合模型改善需求預測｜國立臺灣科技大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	李孟珊 Meng-Shan Lee
論文名稱：	運用基於機器學習演算法之混合模型改善需求預測 Improving Demand Forecasting with Hybrid Forecast Model Using Machine Learning Algorithm
指導教授：	呂志豪 Shih-Hao Lu
口試委員:	曾盛恕 Seng-Su Tsang 郭人介 Ren-Jieh Kuo
學位類別：	碩士 Master
系所名稱：	管理學院 - 企業管理系 Department of Business Administration
論文出版年：	2020
畢業學年度：	108
語文別：	英文
論文頁數：	42
中文關鍵詞：	需求預測、組合預測、機器學習、極限梯度提升、支持向量迴歸
外文關鍵詞：	Demand Forecasting, Combined Forecasting, Machine Learning, eXtreme Gradient Boosting, Support Vector Regression
相關次數：	點閱：544 下載：14
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

上一筆

需求預測在供應鏈管理中扮演至關重要的角色，越來越多的零售商渴望更快速地做出高品質的決策，而隨著近年人工智慧與機器學習風潮興起，業界開始關注如何運用人工智慧與機器學習建立準確且自動化的需求預測系統。本研究的目的是透過機器學習演算法開發更精準的自動化需求預測模型，而本研究提出之模型特點為運用 2 種機器學習模型（eXtreme Gradient Boosting 與 Support Vector Regression）來改善 2 種指數平滑方法（Holt’s Exponential Smoothing 與 Winter’s Exponential Smoothing）的預測結果。透過為每件商品自動地選擇其最佳模型、合適的模型特徵以及最佳參數，本研究提出之模型不僅可以提供每件商品最佳的需求預測模型條件，且可以提供更加精準的預測結果。

在預測績效方面，本研究針對臺灣某一大型零售商的102件傢俱類商品進行需求預測，結果表明本研究提出之模型最佳結果的準確性（sMAPE）高達 93.77％；此外，與純指數平滑模型相比，該模型的預測誤差（sMAPE）降低了 46.47％（從 11.64％降低到 6.23％）。研究結果不僅表明機器學習模型可以顯著改善指數平滑方法的預測結果，且指出為每件商品設計客製化需求預測模型是企業在需求預測方面需要關注的議題。

Demand forecasting plays a vital role in supply chain management. More and
more retailers eager to produce high quality decisions more quickly, which leads to the need of a more accurate “automation” demand forecasting system. The objective of this paper is to develop a more accurate automation demand forecasting model by taking advantage of machine learning algorithms. The proposed model is designed to utilize 2 machine learning models (eXtreme Gradient Boosting and Support Vector Regression) to improve forecast results of 2 Exponential Smoothing methods (Holt’s Exponential Smoothing and Winter’s Exponential Smoothing). By automatically selecting best models, appropriate features and optimal parameters for each product, the proposed model can not only offer each product customized conditions for demand forecast, but a forecast result with high accuracy.

Regarding the forecast performance, 102 furniture items of a major retailer in
Taiwan are applied to the proposed model and the accuracy (sMAPE) of the best
result achieves 93.77%. Additionally, compared to pure Exponential Smoothing
models, forecast errors (sMAPE) of the proposed model decreases 46.47% (from
11.64% to 6.23%). Findings not only reveal that machine learning models can
significantly improve the results of Exponential Smoothing methods, but indicate
that designing a customized demand forecast model for each product can lead to a
better outcome in forecasting demand.

摘要............................................................................I
ABSTRACT.......................................................................II
TABLE OF CONTENTS ............................................................III
LIST OF TABLES .................................................................V
LIST OF FIGURES ...............................................................VI
CHAPTER 1  INTRODUCTION ........................................................1
1.1 Research Background ........................................................1
1.2 Research Objective .........................................................2
1.3 Research Scope .............................................................3
CHAPTER 2  LITERATURE REVIEW ...................................................5
2.1 Traditional Model ..........................................................5
2.2 Machine Learning Model .....................................................6
2.3 Hybrid Model ...............................................................7
CHAPTER 3  METHODOLOGY .........................................................9
3.1 Business Understanding .....................................................9
3.2 Data Understanding ........................................................10
3.3 Data Preparation ..........................................................12
3.4 Modeling ..................................................................13
3.4.1 Hybrid Model Development ................................................13
3.4.2 Trend-Corrected Exponential Smoothing (Holt’s Model) ....................14
3.4.3 Trend- and Seasonality-Corrected Exponential Smoothing (Winter’s Model)..15
3.4.4 eXtreme Gradient Boosting (XGBoost) .....................................16
3.4.5 Support Vector Regression (SVR) .........................................16
3.5 Evaluation ................................................................17
CHAPTER 4  RESULTS ............................................................19
4.1 Overall Time Series .......................................................19
4.2 Each Time Series ..........................................................25
4.2.1 Best Model Combination ..................................................25
4.2.2 Feature Combination .....................................................27
CHAPTER 5  CONCLUSIONS AND FUTURE RESEARCH ....................................30
5.1 Conclusions ...............................................................30
5.2 Future Research ...........................................................31
REFERENCES.....................................................................33 
                                

SAS 臺灣，2019。讓 AI 入魂，精準預測你的需求與庫存！TechOrange 科技報橘。取自:
https://buzzorange.com/techorange/2019/02/13/ai-predict-demand-and-inventory/
Aburto, L. and Weber, R., 2007. Improved supply chain management based on hybrid
demand forecasts, Applied Soft Computing, 7, 136‐44.
Anusha, S. L., Alok, S. and Shaik, A., 2014. Demand Forecasting for the Indian
Pharmaceutical Retail: A Case Study, Journal of Supply Chain Management Systems, 3(2), 564.
Babu C. N. and Reddy B. E., 2014. A moving-average filter based hybrid ARIMA–
ANN model for forecasting time series data, Applied Soft Computing, 23(10), 27–
38.
Bala, P. K., 2012. Improving inventory performance with clustering based demand
forecasts, Journal of Modelling in Management, 7(1), 23-37.
Bousqaoui, H., Achchab, S. and Tikito, K., 2017. Machine learning applications in
supply chains: an emphasis on neural network applications, Proceedings of 2017
3rd International Conference of Cloud Computing Technologies and Applications
(CloudTech), 1-7.
Brea C., Anderson J., Bartling R., Mayo R. and Mueller F., 2018. Demand Forecasting with Advanced Analytics, retrieved from https://www.bain.com/insights/demand-forecasting-with-advanced-analytics/
Brożyna, J., Mentel, G. and Szetela, B., 2016. A Mid-Term Forecast of Maximum Demand for Electricity in Poland, Montenegrin Journal of Economics, 12(2), 73-88.
Chen, T. and Guestrin, C., 2016. Xgboost: A scalable tree boosting system, Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794.
Chu, C. -W. and Zhang, G. P., 2003. A comparative study of linear and nonlinear
models for aggregate retail sales forecasting, Int. J. Production Economics, 86,
217-231.
Cortes, C. and Vapnik, V., 1995. Support-vector networks, Machine Learning, 20, 273–297.
Holt, C.C., 1957. Forecasting seasonals and trends by exponentially weighted moving averages, International Journal of Forecasting, 20, 5-10.
Hyndman, R. J., 2006. Another look at forecast-accuracy metrics for intermittent
demand, Foresight: The International Journal of Applied Forecasting, 4, 43–46.
Jain, A., Menon, M. N. and Chandra, S., 2014. Sales forecasting for retail chains. Working Paper.
Maaß, D., Spruit, M. and Waal, P. D., 2014. Improving short-term demand forecasting for short-lifecycle consumer products with data mining techniques, Decision Analytics, 1(1), 1–17.
Pillo, G. D., Latorre, V., Lucidi, S., and Procacci, E., 2016. An application of support vector machines to sales forecasting under promotions, 4OR Quarterly Journal of the Belgian French & Italian Operations Research Societies, 14, 309-325.
Sayed H. E., Gabbar H. A. and Miyazaki S., 2009. A hybrid statistical genetic-based demand forecasting expert system, Expert Systems with Applications, 36(9),
11662-11670.
Symphony RetailAI, 2018. Study Shows Half of Retail Leaders Are Embracing AI to
Transform the Supply Chain, retrieved from https://www.symphonyretailai.com/study-shows-half-of-retail-leaders-are
embracing-ai-to-transform-the-supply-chain/
Taylor, J. W., 2010. Multi-item sales forecasting with total and split exponential smoothing, Journal of the Operational Research Society, 62, 555–563.
Tofallis, C., 2014. A better measure of relative prediction accuracy for model selection and model estimation, Journal of the Operational Research Society, 66(8), 1352-1362.
Winters, P. R., 1960. Forecasting sales by exponentially weighted moving averages, Management Science, 6, 324-342.
Zhang G. P., 2001. Time series forecasting using a hybrid ARIMA and neural network model, Neurocomputing, 50(0), 159-175.

簡易檢索 / 詳目顯示

相關論文