tsantas
13-07-2015, 18:34
Την Τετάρτη 15 - 7 - 2015 και ώρα 12:30 στο υπολογιστικό κέντρο του πρώην Γενικού Τμήματος θα δοθεί ομιλία από τον Prof Juan Manuel Rodriguez Diaz του Πανεπιστημίου της Salamanca, Spain, με τίτλο "Design of experiments principles and applications"
Abstract
A high executive of the Toyota Motor Company said once that 'An engineer that does not know about design of experiments is not an engineer'. Well, maybe the statement is a bit too much, but what is true is that a well-designed experiment can provide more information without increasing the budget (number of samples, time,...). The objective of an experiment is to learn about a process, for which some trials are done under different conditions in order to find out how these changes of conditions affect to the outcomes of the experiment. A careful choice of the set of conditions where to take the samples is essential for getting more or less information about the unknown characteristics of the process, and making this choice is designing an experiment. In the talk, some examples of well and bad designed experiments are shown and discussed, and the main statistical tools for this purpose are reviewed.
Abstract
A high executive of the Toyota Motor Company said once that 'An engineer that does not know about design of experiments is not an engineer'. Well, maybe the statement is a bit too much, but what is true is that a well-designed experiment can provide more information without increasing the budget (number of samples, time,...). The objective of an experiment is to learn about a process, for which some trials are done under different conditions in order to find out how these changes of conditions affect to the outcomes of the experiment. A careful choice of the set of conditions where to take the samples is essential for getting more or less information about the unknown characteristics of the process, and making this choice is designing an experiment. In the talk, some examples of well and bad designed experiments are shown and discussed, and the main statistical tools for this purpose are reviewed.