University of the Aegean
Διαγράμματα ελέγχου για υποδείγματα διακριτών χρονολογικών σειρών
Abstract
dc:descriptionThe topic of this PhD dissertation is the development of memory-type control charts for monitoring processes that are modeled according to a time series model with integer values. Specifically, the focus is on the study and development of control charts for monitoring BAR(1), BBAR(1) and BINARCH(1) processes. The first model is commonly used to describe binomial data (e.g. number of non-conforming items) with first-order autocorrelation. Moreover, the BBAR(1) and BINARCH(1) models are generalizations of the BAR(1) model and thus appropriate for cases where there is evidence that there is extra-binomial variation in the data. For all the three models, main properties are provided, followed by the construction of the suggested control charts that can be used for the monitoring of the corresponding processes. For the case of BAR(1) processes, one-sided EWMA control charts are proposed and studied. In addition, double EWMA (DEWMA) charts are studied in comparison with the ordinary EWMA charts, in attempt to assess whether the more complex DEWMA chart has better performance or not than its competitor (the EWMA chart). The results from this comparative study are given in the case of BAR(1) process. For the case of BBAR(1) processes, one-sided EWMA charts are also studied. Three different types of one-sided charts (Shewhart, EWMA and CUSUM) are proposed and studied for the monitoring of BINARCH(1) processes whereas the case of two-sided Shewhart and EWMA charts is also discussed. Aspects of the statistical design for all the considered charts, for each of the three processes, are discussed and empirical rules for the selection of the proper values of the parameters of each chart are given. The Markov chain method as well as Monte Carlo simulation are used for the evaluation of the performance of each chart. Finally, numerical comparisons (in terms of average run length) are given for all the considered charts, for the three different processes (BAR(1), BBAR(1) and BINARCH(1)), separately. In addition, the practical implementation of the proposed charts in statistical process monitoring problems is discussed in detail by applying them in real data that are available from three areas of applied science (epidemiology, economics and computer network management).
Degree
thesis:*- Grantor dc:publisher
- University of the Aegean
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Anastasopoulou, Maria
- Αναστασοπούλου, Μαρία
Subjects
dc:subject × 28- Αυτοπαλίνδρομα υποδείγματα
- Διωνυμική κατανομή
- Διωνυμικός τελεστής λέπτυνσης
- Μέσο μήκος ροής
- Στατιστικός έλεγχος διεργασιών
- Διαγράμματα ελέγχου τύπου Shewhart
- Διαγράμματα ελέγχου τύπου EWMA
- Διαγράμματα ελέγχου τύπου DEWMA
- Διαγράμματα ελέγχου τύπου CUSUM
- Πληθωρισμός Ευρωζώνης
- Average run length
- Binomial AR(1) model
- Beta-BaR(1) model
- BINARCH(1) model
- Extra binomial variation
- Rounding operation
- s-EWMA chart
- Statistical process control
- Eurozone inflation data
- Attributes control charts
- Integer-valued time series
- Statistical process monitoring
- Μαθηματικά
- Φυσικές Επιστήμες
- Στατιστική και Πιθανότητες
- Mathematics
- Natural Sciences
- Statistics and Probability
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/54124