Pattern recognition in the Early Warning generation process
| dc.contributor.author | Larsson, Elias | |
| dc.contributor.department | Chalmers tekniska högskola / Institutionen för teknikens ekonomi och organisation | sv |
| dc.contributor.examiner | Bohlin, Erik | |
| dc.contributor.supervisor | Bohlin, Erik | |
| dc.date.accessioned | 2022-03-23T07:27:39Z | |
| dc.date.available | 2022-03-23T07:27:39Z | |
| dc.date.issued | 2022 | sv |
| dc.date.submitted | 2020 | |
| dc.description.abstract | This paper covers an analytical study of Early Warnings - a data object that is automatically generated when a customer order is created. Early Warnigns are then continuously created any time there is a change in expected delivery. Every day, a huge amount of Early Warnings is generated. The amount of data makes it difficult to evaluate and predict changes. A recognition of patterns in Early Warnings could support the planning process and also reduce uncertainty or stress regarding the huge wave of data that daily washes ashore. The study has been conducted by mainly considering the aspects of lead time, delivery date, Early Warning generation date and quantity. By alinging collected data based on the delivery date, a new perspective of the analysis was acheived. Instead of considering dd.mm.yy or week X of the lead time, the data points are sorted as "delivery date - X weeks of lead time". This has been referre to as Temporal variable analysis. Additionally, a grading of Early Warnings was set up to evaluate the accuracy of each update compared to final delivery. The results were presented in color-based scales and in compiled graphs for easy overview and analysis. The five main findings of the paper are that; (1) there is in fact consequent patterns in the analysis set up. (2) The utilization of several order lines, mainly due to partial consignments, strongly increase the number of Early Warnings generated. (3) Expedited delivery dates appear during all parts of the lead time, contrary to initial beliefs. (4) The total sum of shifts of Early Warnings is largely negative, this is due to the increased number of Early Warnings at the late stages of lead time that coincide with the gradually negative values of Early Warnings. (5) The finding can be visualized in a model that is easy to scale up for more empirical results. The main findings for the researcher can be summarized as; - Expected delivery is expedited for orders until a certain lead time until final delivery. After this point, the majority of Early Warnings are postponements of the expected delivery date. - More partial deliveries for an order strongly increases the number of Early Warnings generated and appears to lower the overall accuracy of the Early Warning compared to final delivery. | sv |
| dc.identifier.uri | https://hdl.handle.net/20.500.12380/304546 | |
| dc.language.iso | eng | sv |
| dc.setspec.uppsok | Technology | |
| dc.subject | Early Warning, temporal variable analysis | sv |
| dc.title | Pattern recognition in the Early Warning generation process | sv |
| dc.type.degree | Examensarbete för masterexamen | sv |
| dc.type.uppsok | H | |
| local.programme | Quality and operations management (MPQOM), MSc |