Browsing by Author "Jalloh, Rashid"

Now showing 1 - 2 of 2

Synthesizing a predictive analytics methodology for quality management
(2023-05-15) Jalloh, Rashid
Insinööritieteiden korkeakoulu | Master's thesis
As advancements in sensor fusion and distributed computing among other technological advancements are ushering industry into a fourth revolution, similarly, advancements in predictive and prescriptive analytics are bringing about a fourth revolution in quality management, coined Quality 4.0. Especially when considering products with long lifecycles such as elevators in the case of thesis, predictive analytics enables quality management to act on quality issues even before they occur, saving resources and further expanding the products longevity in service. Despite these benefits, organizations are struggling to take advantage of predictive analytics. One major issue is the lack of standardized ways of running predictive analytics projects, leading organizations to develop ad-hoc methodologies. For two centuries Cross-Industry Standard Processes for Data Mining (CRISP-DM) has been the most widely used methodology for predictive analytics tasks, with many opting to base their domain specific methodologies on CRISP-DM. In this thesis a predictive analytics methodology is synthesized based on CRISP-DM that incorporates familiar tools and ways of working for quality management, such that simple predictive analytics projects can be run without deep understanding of data science or machine learning. For validation of the methodology, a proof-of-concept machine learning application was developed step-by-step for an elevator quality questionnaire. The final model achieved a prediction accuracy of over 60% and a false-alarm rate of below 33%, which were defined as the success criteria for the application. Though the criteria were quite modest, the simple application and the methodology serve as a suitable starting point for running predictive analytics projects in quality management, that with more experience and resources can lead to more advanced analytics with deep learning and predictive maintenance.
Yhteistyörobotit ihmisten työkaverina tuotantoympäristössä
(2020-11-29) Jalloh, Rashid
Insinööritieteiden korkeakoulu | Bachelor's thesis