Domain-invariant partial least squares regression (di-PLS)

[B-Analytics]

I am the main author of "Domain adaptation for regression under Beer–Lambert’s law" and creator of the (unsupervised) domain adaptation extension of partial least squares regression called di-PLS that has been very useful in applications in analytical chemistry and chemometrics. What sets our method apart from most others proposed so far is interpretability. You can actually inspect the domain-invariant representations through the corresponding weights, loadings, and regression coefficients. As discussed with @tgnassou I will start adding the corresponding class (with scikit-learn compatible API) and function from diPLSlib (https://github.com/B-Analytics/diPLSlib). Thanks for the great project and work, and thanks for having me!

Best wishes
Ramin

Reference

Nikzad-Langerodi, R., Zellinger, W., Lughofer, E., & Saminger-Platz, S. (2018). Domain-invariant partial-least-squares regression. Analytical chemistry, 90(11), 6693-6701.

Nikzad-Langerodi, R., Zellinger, W., Saminger-Platz, S., & Moser, B. A. (2020). Domain adaptation for regression under Beer–Lambert’s law. Knowledge-Based Systems, 210, 106447.

[rflamary]

Welcome @B-Analytics

Feel free to ask if you have questions, we have done a lot of work for DA classification but few methods can handle regression so this will be a test for our API.

[tgnassou]

Welcome @B-Analytics

Thanks for joining the project! I'll soon make a contributor guide for you;)

[tgnassou]

@B-Analytics You can find the contributor guide here!

[B-Analytics]

Great, thanks!