# Non-linear regression with a neural network

I have to perform a regression on three curves as shown in the following plot. Here, accA (y-axis) is the dependent variable, and w (x-axis) is the independent variable. The sum of the three curves adds up to 1.

To perform regression for the three curves, I would like to use a neural network. What architecture/set-up should I use for this task?

• Hi and welcome to AI SE! Just to clarify, what exactly do you want to predict here? Do you want to predict the value for each of the 3 curves individually? – nbro Mar 24 '20 at 20:05
• Yes, exactly. I would like to fit three curves individually, basically, but in a way that the sums of their ouput always adds up to 1, for each choice of the variable w. I would like to use a neural network, instead of performing a regression manually. – Fq00 Mar 25 '20 at 8:47
• @Fq00 It looks like you want to do time series prediction maybe? So the real goal is to predict the next value of of the 3 accA functions, given the past values of accA (and maybe w)? – John Doucette Mar 26 '20 at 0:31