Dear Sunita Kesar
A lack-of-fit error significantly more significant than the pure error indicates that something remains in the residuals that a more appropriate model can remove. If you see a significant lack-of-fit (Prob>F value 0.10 or more minor), don't use the model as a predictor of the response.
For the best model, a p-value of a model should be significant (p0.05)
If lack of fit is significant, then these three methods should be tried: 1- Central points should be repeated 2- Higher grade models should be selected 3-The transfer function must be used for the data.
I suggest reading this paper
Conference Paper Application of RSM modelling to a synthetic Gemini surfactan...
Best regards,
Thank you, dear Sangat Naik for your answer...but my query was also to know the reason that if there is no Lack of fit and pure error occurs in a developed model..is it a good model..or not?
Dear Sunita Kesar
I suggest you read the following articles so you get explicit knowledge.
https://online.stat.psu.edu/stat462/node/113/
https://online.stat.psu.edu/stat501/lesson/2/2.11
https://www.theopeneducator.com/doe/Regression/lack-of-fit-test
https://math.unm.edu/~fletcher/SUPER/chap8.pdf
https://support.minitab.com/en-us/minitab/21/help-and-how-to/statistical-modeling/regression/supporting-topics/regression-models/lack-of-fit-and-lack-of-fit-tests/
https://redox-college.s3.ap-south-1.amazonaws.com/kmc/2020/Apr/03/mLCfckXwkRHV2K0r3ZIi.pdf
If you have any queries feel free to contact me regarding the research
Best regards
Thank you for answering @Omidali Najafi. In support of your answer, can you please refer any relevant literature, plzzz?
- Badges
- Science topic