Is it possible to use some form of curve identifcation algorithm or model in R to identify a curve based on data from three different treatments?

Question

I was wondering if it is possible to train a model or algorithmn to identify a curve shape based on experimental data.

Say we have 3 treatments 0 mM sugar, 1 mM sugar and 10 mM sugar and three different bacterial strains: A (A.10.csv, A.1.csv, A.0.csv), B (B.10.csv, B.1.csv, B.0.csv) & C (C.10.csv, C.1.csv, C.0.csv).

A is chemotactic and swim outwards. B is non-chemotactic swims randomly and C is a mixture.

I'd then run the model/ learning algorithm on (test.10.csv, test.1.csv, test.0.csv) and it could suggest which strain it was most similar to ?

Here is the data https://drive.google.com/file/d/1k8-q3RhsVU08sMh5_spSCadfJPUUkR2b/view?usp=sharing

Answer 1

If your model that you have fit to the experimental data can make probabilistic predictions under the three different possible strains, then you can evaluate the log pointwise predictive density of the observed data new under the three different models. You could then compare this performance (see e.g. here).