0
$\begingroup$

I've got a bunch of data with integer values for let's say variables $X_1, X_2, X_3$ and an also integer value $Y$ for each set of $X_{1-3}$. My goal would be to find something like a formula that would show the mathematical operations between the variables $X_{1-3}$ to get the value of $Y$, with the help of computer software. Kind of like reverse-engineering a function into a formula to be able to calculate the value for any $X_{1-3}$ values, or even plot it.

Correct me if I'm wrong, but as far as I understand this process is called interpolation for points in between and extrapolation for points outside my data range. However, since with that I have zero experience, I cannot be sure, especially considering almost every article about those already had a certain function to interpolate, and didn't work with a data set.

So can anyone point a newbie like me in the right direction, and perhaps suggest some software for doing that? My idea was MATLAB, since I have access to a student's licence of it, but if it can't do it I'm open for other suggestions as well.

EDIT: Some additional info that I forgot to mention, and may help: all the variables have positive sign, and may have constant multipliers (a*X1, etc...).

PS: never talked about maths in English before so apologies if something is incorrect :)

New contributor
Dave is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
$\endgroup$
  • $\begingroup$ Since you have discrete integer values that you're trying to predict, I would call this, in the language of machine learning, a classification problem. The standard go-to model for classification is logistic regression, though it might not give you the best results. It's probably the best-understood algorithm. $\endgroup$ – Adrian Keister Aug 12 at 14:43
  • $\begingroup$ I see, and thank you for getting that straight. I would assume in case of machine learning it would take a lot more data than what I currently have. I have around 100 sets of measurements, but from what I've heard these AI-based solutions require tens of thousands, is that correct? $\endgroup$ – Dave Aug 14 at 6:53
  • $\begingroup$ Well, that depends on the data, actually! Sometimes, there comes a point where more data doesn't help all that much. And over-fitting is a constant problem. Definitely recommend logistic regression to start you off. $\endgroup$ – Adrian Keister Aug 14 at 13:40

Your Answer

Dave is a new contributor. Be nice, and check out our Code of Conduct.

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.