1

The control chart usually requires that the observed values meet the conditions of independent and same distribution, but the sample data is a time series with autocorrelation. How should the data be processed to eliminate the autocorrelation

xin liu
  • 11
  • Please, can you give more information on the data? what they represent for example? Why do you think they are correlated? etc...? – lulufofo Jun 06 '23 at 11:58
  • For example, when I obtain the vibration signal of the bearing for monitoring, the data at time t and time T-1 are often correlated, they are not independent, and the control chart based on the variable point model requires the sample data to be independently and equally distributed. My problem is how to process the sensor data and eliminate the correlation between them – xin liu Jun 06 '23 at 12:07
  • If you know how to do simulations (for Bootstrap), it can be worth to read the paper at this link https://www.jstor.org/stable/2291598 . Unfortunately I do not have access to it anymore. – lulufofo Jun 06 '23 at 13:23
  • Welcome to CV, liu. I would like to suggest that a more constructive way to ask this question would be to focus on understanding and accommodating the autocorrelation rather than "eliminating" it. At the same time, it will be difficult to find any objective answer without first knowing what you hope to accomplish by control charting such a time series. Could you explain that? – whuber Jun 06 '23 at 16:27

0 Answers0