I offer consultancy in statistics, biostatistics, machine learning and methods comparison. If you have a problem you can't solve and if you can get to find me you might hire me.
I have resurrected some old methods for Deming's robust regression (M- and MM- or Huber and Tukey if you prefer). I proposed a single test based on the Mahalanobis distance (MD) to replace the typical slope and intercept confidence interval tests performed separately and without FWER. Finally, I developed a Bayesian method for Deming regression that can be run in robust mode and also has models suitable for heteroschedastic data. I maintain two packages on CRAN whose sources can be found below:
- mcrPioda: this is a fork of the mcr package with additional M- and MM-Deming regressions and the function for the MD test.
- rstanbdp: is the package for dealing with Deming regressions in Bayesian mode. Robust methods and methods for heteroscedastic data are available.
Here is an interesting read from 2021 on method comparison and Passing Bablok bias
Here my new advances for method comparison
I have a PhD in chemistry (like Wilcoxon and Tukey) and have previously worked on:
- homogeneous enantioselective catalysis
- reaction kinetics
- 3D structure determination of proteins via NMR
- electroconductive graphite based dispersions
