I'm working on a project involving around 90 3D-printed cubic samples with different structures. After conducting a compression test, I obtained stress-strain curves with 700 data points for each sample, resulting in a total of 63,000 values. Here's a snapshot of the original data for two samples:
My goal is to implement the best-trained model that allows users to input stress and strain and obtain predictions for the best Lattice Structure, X, Y, Z, and Thickness.
To simplify the dataset, I've categorized strain into specific ranges (0 to 12.5) and defined the min, average, and max stress for each range, resulting in a more manageable dataset with 450 rows, as shown below:
While I have multiple rows with the same targets, similar to this question, I'm facing a regression problem with multiple targets. The suggestion in this answer to concatenate rows won't work for me, as I need separate inputs (as in the last table) for integrating the ML model into a web application.
Since building an ML model with the same targets may not yield accurate results, I'm seeking advice on how to address this regression issue with multiple targets and still maintain the necessary input structure for my web application integration.
Any insights or guidance would be greatly appreciated.