Curve fitting of these two variables would depend on an approximate order of the relationship between the two variables. Let us say if they have a second-order non-linear relationship, then most curve fitting tools find the coefficients of the second-order polynomial which best fits the 80 points you have.

MS Excel or Matlab both can be used (for the above problem Excel would do).

If your curve needs a combination of nonlinear functions to represent it, then you would have to do a minimum mean square error based fit (Excel has solver for this too).

useful link:

https://www.youtube.com/watch?v=n4XYABtC3Dk&ab_channel=ignite.byu.edu

https://www.youtube.com/watch?v=Ewp5CF5ba_w&t=71s&ab_channel=TaylorSparks

Without doing data analysis can't say anything particularly, it depends on data.

Start from linear regression. Then try some Feature Engineering.

Then go to second or third order polynomials + feature engineering.

Then try to build tree models visualize it or then go to random forest or then XGBoost or something. Then try a simple neural network.

You can try the Multivariate Linear Regression Model,

You can look into K Matrix and P Matrix Calibration Models in this regard, the following book discusses in depth the mathematical development of these models:

" Mathematical Analysis of Spectral Orthogonality"

by Kalivas and Lang