Μη παραμετρικά διαγράμματα ελέγχου με προοδευτικά επικαιροποιημένο δείγμα Φάσης Ι για μοντέλα αλλαγής σημείου
Nonparametric control charts based on a sequentially up-to-date Phase I sample under change-point models
View/
Keywords
Μη παραμετρικά διαγράμματα ; Διατεταγμένες παρατηρήσεις ; Προοδευτική επικαιροποίηση διαγράμματος ; Προσομοίωση Monte Carlo ; Στατιστικός έλεγχος ποιότητας ; Non-parametric diagrams ; Ordered observations ; Progressive diagram updating ; Monte Carlo simulation ; Statistical quality controlAbstract
The main scope of this thesis is to focus on the design, simulation and evaluation of non-parametric diagrams of ordered observations Precedence, aiming at the transition from traditional static models to modern approaches through progressive sample updating. The purpose of this research effort is to address the issue of requiring a very large sample in Phase I. Through extensive Monte Carlo simulations in the R environment, a static model is simulated, which confirms the property of the free distribution, keeping the under-control Mean Flow Length constant, regardless of the distribution, while at the same time its comparison with the Median Flow Length highlighted the strong skewness of the distribution in terms of response time, urging the use of the Median Length as a more reliable measure at small displacements. In the last chapter of this work, a progressive algorithm was developed, which starts with a small reference sample (trying to overcome the problem of the requirement for a large sample), dynamically updating the control limits. In the context of the above approach, a critical aspect was revealed, capturing the issue of the masking effect, that is, the case in which this progressive system incorporates slightly defective samples, causing an unintentional widening of the limits, adding a time delay in the context of error detection compared to the static model. On the contrary, at larger displacement values, error detection became immediate with the performances of the two diagrams being approximately the same.
Postgraduate Studies Programme
Εφαρμοσμένη ΣτατιστικήDepartment
Σχολή Χρηματοοικονομικής και Στατιστικής. Τμήμα Στατιστικής και Ασφαλιστικής ΕπιστήμηςNumber of pages
102Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα