University of Twente Student Theses
Developing a Labour Cost Prediction Algorithm for an Industrial Packaging Configurator
Eijkholt, S.M. (2024) Developing a Labour Cost Prediction Algorithm for an Industrial Packaging Configurator.
PDF
3MB |
Abstract: | Meilink is a Dutch family business in high-tech industrial packaging. The firm encountered a problem in predicting labour cost for configurable products. The configurator accurately calculates required resources but cannot predict labour costs, therefore, it is complex to provide accurate quotes. We evaluated nine supervised machine learning techniques to utilize historical data to predict labour costs of future products. We split the dataset in 80% training data and 20% test data. We evaluated the performance of each of the techniques with a composite score of Mean Squared Error and Akaike Information Criterion, to assign a suitable technique to each of the six considered configurable products. A prediction with the actual value within its 95% confidence interval classifies as sufficient. We denote the algorithms accuracy as the percentage of sufficient predictions, assess generalizability with K-fold model validation, and exclude outliers using the IQR method. We achieved an overall improvement of 21.74% at a 5% level of significance. The implications of our improvement affect many facets of the organisation: customers are more satisfied as the quotations represent invoices, scheduling precision increases, and it is an overall practical solution for the sales department. |
Item Type: | Essay (Master) |
Clients: | Meilink B.V., Borculo, Netherlands |
Faculty: | BMS: Behavioural, Management and Social Sciences |
Subject: | 50 technical science in general, 85 business administration, organizational science |
Programme: | Industrial Engineering and Management MSc (60029) |
Link to this item: | https://purl.utwente.nl/essays/99831 |
Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Repository Staff Only: item control page