Machine learning methods in seismic facies analysis problems

The principles of machine learning and its use in solving seismic facies analysis problems are considered. The issues of the efficiency of the obtained solutions at a qualitative level, as well as significant aspects influencing their efficiency are discussed. The latter include: data quality, groups of attributes used, features of clustering algorithms. As an example, the results obtained for horizons related to the Bobrikovian–Tournasian strata of the Lower Carboniferous age in the southwestern part of the Orenburg region are given.