Sorry for my English.

Dataset

There is dataset which contains files which describe scheme:

sample #1.txt

3103686, 2590304, 2022230, 838696
5530360, 1916721, 2022230, 430823
3103686, 3807071, 2022230, 430823
5705725, 4022485, 2022230, 975943
8043677, 3697167, 2022230, 430823
8043677, 2761756, 2022230, 430823

sample #2.txt

2994926, 3072910, 2022230, 1752477
7396944, 3072911, 2022230, 1752476
2994926, 1981531, 5573177, 558310

Each row is rectangle element (on scheme) feature vector (x, y, width, height).

Data to predict

I need train a model which can predict for such input data

input.txt

3313321, 3259181, 2022230, 558310
7039277, 3454335, 2022230, 558310
5253403, 4207799, 2022231, 558310
4073770, 2445894, 2022230, 558310
6569923, 2445894, 2022230, 558310

similar scheme.

For example, in the above example for input.txt prediction would be quite if model say that sample #1 most similar for input scheme.

Question

Which algorithm from ML.NET should I use to solve my task? Of cause I do not expect complete solution, just put me right way.
I have a little bit sub-questions to clarify my problem:

• How preparing dataset to train: by feature describe or matrix?

• Before some classifier should I clustering data?