0

I have two classes of points "success" (1) and "failure" (0) in 2-dim XY-space, I am trying to find the best possible point (or region) of space where the success is highly likely. Meaning that if I take a new point that lands around that point (or region) is likely to be succes.

Is there any ideas how to choose the best success point?

enter image description here

Artashes
  • 102
  • 1
  • 9
  • What did you try so far? – accdias Jul 12 '22 at 15:42
  • In order to answer which point is most likely, you need to model the probabilities of future points somehow, and which model you choose will change the point which is most likely. Which model is best is a somewhat subjective question, but I think it would be easier to answer if you provided more detail about what you're modelling. – Nick ODell Jul 12 '22 at 15:54
  • @accdias, weighted-kNN based algoritmses, center of gravity of succes points, mean-shift like algorithms. Each method gives completely different results. I also tried the mean of all methods and for the moment I stick with that. Do you have any methode suggestion? – Artashes Jul 12 '22 at 15:56
  • Show us the code. – accdias Jul 12 '22 at 15:57
  • @NickODell, we have a phenomena with 2 outcomes, the probabilty distributions are not provided, nor a physical model. All we have is the raw data. And the goal is to find the most probable succes point or region of space. – Artashes Jul 12 '22 at 16:04
  • @accdias, it is a DataFrame with 3 columns : (X,Y,Respons) – Artashes Jul 12 '22 at 16:06
  • By "find the best possible point (or region) of space where the success is highly likely," do you mean a region where a successful point is likely to land, or a region where, assuming a point lands there, it is likely to be successful? Because those are slightly different questions. – Nick ODell Jul 12 '22 at 17:18
  • One other suggestion I would give for how to model it is LDA or QDA classification. – Nick ODell Jul 12 '22 at 17:21
  • @NickODell, assuming a new point that lands there is likely to be successful. We want to avoid the failure. We have to fix X and Y parameters that allowd to obtain only sucess. (We want to avoid the failure points as much as possible) – Artashes Jul 13 '22 at 07:45

0 Answers0