Cluster Regression Can Be Useful…
Event name: Geomodel 2018
Session: Машинное обучение и многомерный анализ данных
Publication date: 10 September 2018
Info: Extended abstract, PDF ( 858.57Kb )
Price: € 20
In petrophysics and geophysics several algorithms of clustering (usually for separating facies, seismic facies and lithotypes) and approximation are widely used. This makes it appropriate to search for new algorithms and to identify the area of preference for each of them. The report considers cluster regression and two model problems for the solution of which it is applied. On the data sets of laboratory measurements on the core for one of the fields in Western Siberia (the deposits include the entire range of layers from BV-8 to J2 inclusive), problems of porosity permeability restoration are considered taking into account differentiating variables and porosity in depth and acoustic properties. On the quality of the approximation of the training sample, cluster regression significantly exceeds the multidimensional regression. The area of the preferred application of the algorithm is analyzed.