So, I have a lot of 3D Plot data that are either in the A category or B category. A short truncated sample of one of the data files is this:

ID, X, Y, Z, Energy discharge
0 1.0 -1.0 1.0 0.00871970156553 1
1 3.0 -1.0 1.0 0.0074666937698 1
2 3.0 -1.0 3.0 0.00383590709193 0
3 3.0 -3.0 3.0 0.00433503021393 0
4 5.0 -3.0 3.0 0.0121560751343 1
5 7.0 -3.0 5.0 0.0111660665889 1
6 9.0 -3.0 7.0 0.0106192726346 1
7 11.0 -3.0 9.0 0.0103951518801 1
8 13.0 -3.0 11.0 0.00848731974185 1
9 13.0 -3.0 13.0 0.00898754800867 1
10 15.0 -3.0 13.0 0.00314215753738 0
11 15.0 -3.0 15.0 0.0103174830021 1
12 17.0 -3.0 15.0 0.00150950214803 0

When plotted, the graph will have many points that will form 3D lines if each point is connected. I would like an algorithm where I can give a lot of these data files and say, this is A or this is B depending on the data file and it can possibly find distinctions from A or B (they look very extremely similar from a human standpoint).

Once the program has found possible distinctions, I would like to test it by giving it an unknown A or B file and see if it can determine what category it is.

Note: The data itself may be out of order so trying to create a linear model by connecting each consecutive point in the data file is not possible - you cannot use the ID as well but it should use the Energy discharge when learning. Also, that extra bit at the end of each line (either 1 or 0) is also non-essential and should not be used when learning.

What is the best machine learning algorithm that is capable of finding distinctions after iterating through say hundreds or thousands of these data files and can make predictions on unknown A or B data files to use in this particular case?

  • $\begingroup$ It is hard to say what will be the 'best' machine learning algorithm. What you need to do is figure out how your data behaves- for example is it linearly separable, and start a machine learning pipeline to find a solution which is best for you. It really depends on how hard the problem is, but it is hard to determine for the information provided in your example data set and is not a very specific question. $\endgroup$ – user3494047 Oct 5 '16 at 14:38
  • $\begingroup$ @user3494047 The data points form a linear model if connected but the data that comes in may be random so a linear model (created by drawing a segment point by point) would be incorrect. Right now, I simple have a data file with thousands of points. Is there an algorithm that can simply do analysis based on these points or does it have to be in linear form? What else should I include in this question? How is this not a specific question? $\endgroup$ – Tom Oct 5 '16 at 15:03

Your Answer

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.