Дослідження і розроблення методів і алгоритмів неієрархічної кластеризації

Розроблено і досліджено методи й алгоритми неієрархічної кластеризації, які дають змогу визначити оптимальну початкову кількість кластерів без будь-якої початкової інформації про їхнє розміщення. Розроблені методи і алгоритми досліджено на відомому тестовому наборі Iris.

Developed and studied the methods and non-hierarchical clustering algorithms for determining the optimal initial number of clusters without any background information on the location of the clusters. The methods and algorithms are studied in the famous test set Iris.

Особливості опрацювання даних для ієрархічної кластеризації складних схем

Пропонується гнучкий і універсальний підхід для опису електричних схем та дерева згортання, за яким можна оптимізувати виконання ключових етапів ієрархічної кластеризації.

A flexible and universal approach for presentation of electric circuits and reduction tree, which can optimizes the performance of key stages of hierarchical clustering is proposed.

Алгоритми кластеризації робочого поля з обмеженнями для задачі комівояжера

Описано три підходи до кластеризації робочого поля для задачі комівояжера, що забезпечує поділ множини точок на частини з заданими обмеженнями. Один із відомих алгоритмів використовується для отримання розв’язків в кожному кластері з подальшим зшиванням часткових розв’язків.

Article describes three approaches to clustering set of points of TSP into subsets with given constraints. One of the well-known basic algorithms is used for solutions at every cluster with further joining of partial solutions.

Solutions and approaches analysis for geospatial data clustering to optimize performance and user experience of web maps

In the contemporary epoch, the management and visualization of geospatial information in web browsers have gained substantial importance. Web maps are indispensable tools across various tourism, goods delivery, and ecology sectors. Furthermore, the extensive support of web browsers on diverse devices enhances the accessibility of geospatial data on the web for various users. However, the incessant increment of geospatial information poses new challenges in efficiently displaying data and navigation through these data on web maps.

Intelligent system for clustering users of social networks based on the message sentiment analysis

The main objective of this article is the analysis of the intelligent system for clustering users of social networks based on the messages sentiment analysis. The main goal of this intelligent system is to form a general image of the user of the system by analyzing the sentiment of the data of the user's social networks and their subsequent clustering. An intelligent system was designed, which, using the Identity and Access/Refresh JWT token algorithms, provides fast  and maximally secure registration, authentication and processing of various system user sessions.

Estimation of socio-economic development of districts in Lviv region: the cluster approach

This article provides a cluster analysis of the socio-economic status and development of districts in Lviv region for a period of five years. The indicators that have the highest impact on cluster formation are detected. The author proposes an approach which can serve as a basis for developing effective territory management tools based on statistical analysis of socio-economic indicators.


The number of clustering methods and algorithms were analysed and the peculiarities of their application were singled out. The main advantages of density based clustering methods are the ability to detect free-form clusters of different sizes and resistance to noise and emissions, and the disadvantages include high sensitivity to input parameters, poor class description and unsuitability for large data. The analysis showed that the main problem of all clustering algorithms is their scalability with increasing amount of processed data.

Ієрархічне острівкування енергетичних мереж

The features of using the method of optimal scheme reduction for islanding of power
systems is reviewed. A modified algorithm for series-parallel folding with the formation of
hierarchically nested clusters is offered. The advantages of the algorithm is grounded and its
implementation in the application for islanding of power system is described.

Models and methods for building web recommendation systems

Modern Word Wide Web contains a large number of Web sites and pages in each Web site. Web recommendation system (recommendation system for web pages) are typically implemented on web servers and use the data obtained from the collection viewed web templates (implicit data) or user registration data (explicit data). In article considering methods and algorithms of web recommendation system based on the technology of data mining (web mining).

Application of Cluster Approach to the Development of Industrial Potential: Innovative Policy and Management Support

In modern conditions, the role of industrial potential is increasing, which will ensure the modernization of the national economy. The emergence of crisis phenomena of clusterization in the industrial sector of the state is deepening due to the lack of directions in the management of its development.