A new way to calculate thermal summation constants for forensically useful insects using fuzzy regression
Applied fuzzy regression modeling and interval-based estimation to improve parameter stability in non-linear data-driven models.
Implemented fuzzy regression techniques to derive interval estimates for thermal summation parameters, replacing standard linear approximations. Evaluated model uncertainty by comparing fuzzy-based intervals with conventional regression outputs and visualized performance differences. Built reproducible workflows for interval regression analysis and uncertainty quantification.
This project had a rocky start, but it ultimately turned into a valuable experience that I’m proud of. The initial concepts, while they didn’t pan out as planned, gave me the opportunity to dive deep into fuzzy regression — a challenging but rewarding area of study. Although those early ideas were eventually set aside, they laid the groundwork for the next iteration, which will be published soon. This version leverages advanced methods like finite mixture models, the EM algorithm, and even MCMC algorithms during testing.
At present, you can find a poster in this repository that showcases the canceled version of the project. More details will be released soon, including the abstract, Python scripts, a pre-print, and a GUI program designed for non-technical users.
The current version of the project is avaliable in survival_analysis_em Repo.