Στατιστικός έλεγχος διεργασιών για λογοκριμένα δεδομένα
Statistical process control for censored data

View/ Open
Keywords
Λογοκριμένα δεδομένα ; Στατιστικός έλεγχος ποιότητας ; Shewhart ; Cusum ; EWMAAbstract
This master's thesis focuses on the study and implementation of quality control charts for Type I right censored data, which are frequently encountered in real-world manufacturing process monitoring applications. The work consists of four chapters. The first chapter establishes the fundamental concepts of Statistical Quality Control, placing particular emphasis on Shewhart, CUSUM, and EWMA control charts. Additionally, it explores various types of data censoring mechanisms and presents the form of the likelihood function under censored data. The second chapter provides a comprehensive analysis of Shewhart-type control charts when applied to censored data. This analysis employs the Conditional Expected Value (CEV) approach and investigates its performance across multiple probability distributions, demonstrating the versatility of this methodology. In the third and fourth chapters, the research extends to CUSUM and EWMA methodologies to accommodate censored data environments. These adaptations are specifically designed to enhance the detection capabilities for small or gradual process shifts, which are often challenging to identify using traditional control chart approaches. The theoretical framework is complemented by practical implementations using both real-world datasets and simulated data through the R programming language. These computational studies provide empirical evidence supporting the effectiveness of the examined control charts in identifying out-of-control conditions when data censoring is present.


