Hey, have you ever tried to develop a functional logistic regression model with a quadratic term, only to find that it doesn’t make a significant impact on real datasets? I’m currently facing this challenge, and I’m not alone. While simulations look promising, real-world data just doesn’t seem to benefit from the quadratic term. In fact, sometimes the linear model performs better. 😞
I’ve been searching for datasets where the quadratic term could provide a meaningful improvement, and I stumbled upon the Tecator dataset. This dataset contains the absorbance spectrum of meat samples measured with a spectrometer, and the goal is to predict fat, protein, and moisture content. The quadratic term shows a notable improvement in this case.
But I’m wondering, are there other datasets like Tecator where the quadratic term might shine? I’ve tried audio-related datasets, thinking the quadratic term might highlight certain frequency interactions, but unfortunately, that didn’t work out as expected.
If you have any suggestions or guidance on how to identify promising use cases, I’d love to hear them. Do you know of any real datasets where the quadratic term makes a significant difference?
Let’s explore this together and find some datasets that can help us unlock the potential of functional logistic regression with quadratic terms.