Supply Chain Risk Analysis

Big data exist for supply chains within, for example, enterprise resource planning systems. The data sets are large and complex and include variables linked to, for example, supplier delivery and quality performance, for which observations are generated dynamically through time. Such data can be used to estimate and predict events, such as late delivery of orders or nonconforming parts. Understanding and anticipating the risk of supply chain events allows managers to identify mitigation or control actions to minimise operational disruption.

Computation of key supply risk performance measures can be complicated because we need to appropriately measure the exposure of a supplier to risk. Ongoing work by the supervisor team has developed theoretical methods for estimating, ranking and predicting operational supply risk events. However these methods have been developed under a particular set of stationary assumptions about the probability model which generates the observed data and so are relevant for mature suppliers. There is a need to examine models for new and innovative suppliers. Further, there has been little development of algorithmic rules for optimal data preparation to, for example, determine relevant exposure to risk and this is required because it allows the inference methods to be integrated with the supply chain data management process. This PhD project aims to address these challenges which are important if we are to provide a robust suite of methods that can form a useful decision support tool for operational supply risk management.

To meet our goal, we identify the following four objectives of the research.

1. To characterise the typical patterns observed in real supply operational data sets for a cross-section of high value supply chains using data mining to surface emergent common trends.

2. To develop inference methods to estimate and predict operational supply risk events under assumed non-stationary probability models which are hypothesised to generate the empirical observations typical of those identified.

3. To evaluate the inference methods scientifically through theoretical and empirically controlled studies to verify correctness.

4. To determine and validate useful rules for data preparation that allows the inference methods developed to be directly related to the relevant elements of the big data routinely collected to support supply chain management.

Eligibility
We have an exciting opportunity for PhD research to develop statistical inference methods to support risk informed supply management decisions. Candidates should have competence in probability theory and statistical inference with a motivation to develop novel methods for risk analysis. There are opportunities to engage with industry and to acquire data from supplier databases to both formulate model development and to support validation.

This PhD project requires a highly numerate graduate with skills and interests in computational science. Candidates should have at least a strong Honours degree or equivalent (a strong 2:1 Honours degree, or an undergraduate degree with 3.3 GPA in a 4.0 system), or preferably a Master’s degree in a quantitative discipline such as industrial engineering, operations research, mathematics or computer science (amongst others). Candidates who are not native English speakers will be required to provide evidence for their English skills (such as by IELTS or similar tests that are approved by UKVI, or a degree completed in an English speaking country)

Contact
Informal enquiries to Professor John Quigley ([Email Address Removed]) Department of Management Science.

Applications to: Alison Kerr ([Email Address Removed])