FEATURES OF CONSTRUCTING A CLUSTER LINEAR REGRESSION WITH A COMMON PART
Authors:
1.
Irkutsk State University of Railway Transport
(Doctor of Technical Sciences, Professor; Professor of the Department of Information Systems and Information Security)
Irkutsk, Irkutsk region, Russian Federation
Irkutsk, Irkutsk region, Russian Federation
2.
Irkutsk State Transport University
(Lecturer of the Department of Information Systems and Information Security)
Russian Federation
Russian Federation
3.
Irkutsk State University of Railway Transport
Type:
Article
Pages:
from 142
to 145
Status:
Published
Received:
14.01.2026
Accepted:
02.03.2026
Published:
05.04.2026
Language:
Russian
Keywords:
CLUSTER LINEAR REGRESSION WITH A COMMON PART, METHOD OF LEAST MODULES, LINEAR-BOOLEAN PROGRAMMING PROBLEM, AVERAGE PERCENTAGE ERROR, HYDROELECTRIC POWER STATION, HYDROELECTRIC UNIT
Abstract:
The paper provides a brief overview of publications on various issues of developing methods for constructing cluster models of various types and applications. In particular, the problem of predicting the remaining service life during maintenance of an aircraft engine under various operating conditions is mentioned, which is solved using a self-adaptive approach to dynamic clustering to select useful multimodal data into various clusters, each of which has a constant tendency to degradation. An extension of the well-known k-means algorithm to a multi-criteria structure is considered, which is based on determining the multi-criteria distance based on preferences determined by the decision maker. A summary of the clustering methodology for direct grouping of time series and an automatic recursive search strategy for analyzing energy periodicity in these series is given, as well as a new fuzzy time series model that can interpolate historical data for effective forecasting for the future and, after normalization of the initial data, creates an automatic algorithm to determine the appropriate number of clusters and find fuzzy ones. the relationships of each element in the sequence with the established clusters. A concise description of a new incremental fuzzy cluster ensemble method is presented, based on the idea of combining cluster analysis tasks with classification methods, as well as an approach to group clustering to improve the categorization process when planning workloads in cloud data centers. A method is proposed for identifying the parameters of a cluster linear regression with a common part and grouping variables in the corresponding sets of indexes, suggesting reducing this problem to a linear Boolean programming problem. A cluster and conventional linear model of the operation of a hydroelectric unit of one of the Siberian hydroelectric power plants are constructed. Their brief analysis is carried out.
The paper provides a brief overview of publications on various issues of developing methods for constructing cluster models of various types and applications. In particular, the problem of predicting the remaining service life during maintenance of an aircraft engine under various operating conditions is mentioned, which is solved using a self-adaptive approach to dynamic clustering to select useful multimodal data into various clusters, each of which has a constant tendency to degradation. An extension of the well-known k-means algorithm to a multi-criteria structure is considered, which is based on determining the multi-criteria distance based on preferences determined by the decision maker. A summary of the clustering methodology for direct grouping of time series and an automatic recursive search strategy for analyzing energy periodicity in these series is given, as well as a new fuzzy time series model that can interpolate historical data for effective forecasting for the future and, after normalization of the initial data, creates an automatic algorithm to determine the appropriate number of clusters and find fuzzy ones. the relationships of each element in the sequence with the established clusters. A concise description of a new incremental fuzzy cluster ensemble method is presented, based on the idea of combining cluster analysis tasks with classification methods, as well as an approach to group clustering to improve the categorization process when planning workloads in cloud data centers. A method is proposed for identifying the parameters of a cluster linear regression with a common part and grouping variables in the corresponding sets of indexes, suggesting reducing this problem to a linear Boolean programming problem. A cluster and conventional linear model of the operation of a hydroelectric unit of one of the Siberian hydroelectric power plants are constructed. Their brief analysis is carried out.
Keywords:
CLUSTER LINEAR REGRESSION WITH A COMMON PART, METHOD OF LEAST MODULES, LINEAR-BOOLEAN PROGRAMMING PROBLEM, AVERAGE PERCENTAGE ERROR, HYDROELECTRIC POWER STATION, HYDROELECTRIC UNIT
CLUSTER LINEAR REGRESSION WITH A COMMON PART, METHOD OF LEAST MODULES, LINEAR-BOOLEAN PROGRAMMING PROBLEM, AVERAGE PERCENTAGE ERROR, HYDROELECTRIC POWER STATION, HYDROELECTRIC UNIT
