- Title
- Forecasting in supply chains: the impact of demand volatility in the presence of promotions
- Creator
- Abolghasemi, Mahdi
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2019
- Description
- Higher Doctorate - Doctor of Philosophy (PhD)
- Description
- Demand forecasting is the basis for many managerial decisions in the supply chain. There are many variables such as promotion, market trend, and seasonality that may impact demand and cause volatility. Demand volatility renders demand forecasting a challenging task. Demand volatility is a threat for the supply chain that can cause bullwhip effect (BE), distorts the inventory control, decrease the supply chain performance, and impose extra unnecessary costs to the supply chain. Despite the importance of the demand volatility in supply chain performance, it is often neglected in developing demand forecasting models. In this research, we are concerned with the impact of demand volatility in supply chain forecasting practices. We aim to explore four important forecasting problems in the supply chain context that are caused by demand volatility. We investigate demand forecasting, hierarchical demand forecasting, information sharing, and judgemental forecasting in four main core chapters of this thesis. We develop methods and models for each of them to cope with demand volatility that is mainly caused by promotion. The analysis and results of this research are tied to an empirical case study. The results of this study in Chapter 2 shows that in general demand volatility have a significant impact on the performance of demand forecasting models. Managers need to consider demand volatility as a criterion in developing suitable forecasting models to avoid the negative consequences of an inaccurate forecast. We empirically analyse 844 demand time series that exhibit different levels of volatility and show that the decomposition-based approach are promising when demand is highly volatile. In Chapter 3, we investigate the hierarchical forecasting models. We analyse 63 hierarchical time series where different nodes of the hierarchy have different levels of volatilities. We show that when dealing with these types of hierarchies, it is important to consider an appropriate approach so that we can determine and forecast a group of demand series that are more homogeneous. We show that in volatile series the proportions of each node from the parent node are constantly changing and we must use dynamic models to forecast them. To do so, we use machine learning (ML) models to forecast the proportions of the lower series in the hierarchy and show that ML models are superior to current conventional models. In Chapter 4, we aim to forecast the orders of distribution centres (DCs) when both their orders and demands are volatile. We assess the value of information sharing in the supply chain (in particular, POS data) in order forecasting when the manufacturer has access to both points of sales (POS) data and historical orders of DCs. We use both sources of data separately and show that order-based models are superior to POS-based models to forecast DCs orders. Then, we identify the influential variables that make demands and orders volatile and contribute to the superiority or inferiority of POS-based forecasting models. In Chapter 5, we address the judgemental forecasting problem in the supply chain when experts impose their contextual knowledge to capture the volatility of time series during promotions. We develop an algorithm and propose a simple statistical model that helps experts to structure and quantify their systematic information during judgement. We show that our proposed model outperforms the current practice of judgemental forecasting in the investigated case study.
- Subject
- demand forecasting; supply chains; demand volatility; distribution centres
- Identifier
- http://hdl.handle.net/1959.13/1408493
- Identifier
- uon:35847
- Rights
- Copyright 2019 Mahdi Abolghasemi.
- Language
- eng
- Full Text
- Hits: 2498
- Visitors: 3143
- Downloads: 839
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 21 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 1 MB | Adobe Acrobat PDF | View Details Download |