New Method for Finding Outliers in Nonlinear Regression
Author Information
Author(s): Harvey J Motulsky, Ronald E Brown
Primary Institution: GraphPad Software, Inc.
Hypothesis
Can a new method effectively identify outliers in nonlinear regression analysis?
Conclusion
The ROUT method effectively identifies outliers in nonlinear regression with reasonable power and few false positives.
Supporting Evidence
- The ROUT method detects outliers with an average false discovery rate of less than 1%.
- In simulations, the method identified outliers in 86% of cases when multiple outliers were present.
- The method was shown to be robust against data that follows a Lorentzian distribution.
Takeaway
This study introduces a new way to find outliers when fitting curves to data, which helps make the results more accurate.
Methodology
The study developed the ROUT method, which combines robust nonlinear regression with outlier detection using the false discovery rate.
Potential Biases
The method assumes that outliers are due to experimental errors rather than biological variation.
Limitations
The method may not perform well if more than 30% of the data points are outliers.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website