Using Experimental Design for Calibration Data
Author Information
Author(s): Suzanne Schönkopf, Dominique Guyot
Primary Institution: Camo ASA
Hypothesis
Can experimental design techniques improve the selection of representative calibration data?
Conclusion
The study demonstrates that using experimental design can effectively generate representative calibration data for predictive modeling.
Supporting Evidence
- Experimental design can create structured variation in data.
- Fractional factorial designs require fewer experiments while still providing balanced data.
- Using representative data improves the chances of successful calibration models.
Takeaway
This study shows that by carefully choosing how to experiment, we can get better data to help us make predictions.
Methodology
The study utilized factorial and fractional factorial designs to systematically vary experimental conditions and identify significant factors affecting yield.
Potential Biases
The selection of samples based on experimental design may introduce bias if not all variations are adequately represented.
Limitations
Some experiments could not be performed with the stipulated settings, which may affect the analysis.
Want to read the original?
Access the complete publication on the publisher's website