I have completed the supercritical fluid extraction of my sample. However, after I have fill in the yield values into the design expert, the results suggested that the model is linear even though I have three parameters for the design. Then, the R2 value is too low (lower than 0.5) and the value of adjusted and predicted R2 differs of more than 0.2, so how should I proceed from here as I am clueless on what I need to do in order to improve the design and also to optimize the parameters.
Thank you in advance for answering my queries.
Hi
First: try to implement simple rules for model optimization based on DOE (design of experiments) methodology.
Second: try to implement a more detailed validation methodology for model (obs/pred) rating and evaluation. A quite easy tool could be the LOO method.
See some examples of the LOO procedure implementation for the model rating (obs/pred):
Article Leave-one-out Procedure in the Validation of Elimination Rat...
Article Correlation Between In Silico Physicochemical Characteristic...
Article Relationship Between Drug Structure and Minimal Inhibitory C...
Best regards
Tomasz
do something old fashioned, look at your results... and proceed to model from this, not stats and R2.
Hi Rizal Ali
It depends on the stage of the experiment. If the experiment is almost at the beginning and you have enough budget, you should look at the link https://www.itl.nist.gov/div898/handbook/
Otherwise, you can proceed as Catherine M G C Renard said.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication