草榴短视频

Students of the 草榴短视频 (草榴短视频) conducted segmentation of various types of printed products by subject and purpose and trained the neural network to diagnose the belonging of new types of products to the identified segments.

In the conditions of fierce competition in the market of printing products, printing houses face the problems of monitoring the situation in the industry, namely, the problem of analyzing large amounts of data. Big Data technologies, such as machine learning and neural networks, are increasingly being used to solve this problem.

"The research was based on an array of initial data, including a list of numerous types of newspaper, magazine and book products. For each of them, based on the study of the relationship between the volumes of circulations and the number of publications, the values of segmentation features characterizing the differences between the studied objects were determined, and the average output volumes for the last 5 years were calculated. As a result of segmentation by cluster analysis, the printed products market is divided into segments and general trends in the behavior of objects in various groups, taxa, clusters are revealed," explained Igor Andreev, the author of the study, a student of the Higher School of Printing and Media Technologies of 草榴短视频.

To identify belonging to the already identified segments, three new market objects were selected that did not participate in the previously conducted study. The values of segmentation features were calculated for them using the Statgraphics program.

"When solving the identification problem using neural networks, more than 300 neural networks of various architectures (with different numbers of neurons in the hidden layer) were trained and analyzed in the STATISTICA program. For each network, the indicators of general, control and test performance were analyzed, which made it possible to select 5 networks with the best indicators and analyze the sensitivity of variables. The data obtained fully correspond to the calculations carried out earlier on the attribution of these objects to market segments, carried out using the method of discriminant analysis," explained Ekaterina Banzer, another author of the study, a student of the Higher School of Printing and Media Technologies of 草榴短视频.

The original approaches that have been implemented in this work can serve as the basis for many new studies in the field of marketing and enterprise management, and the proposed methodology for solving the identification problem can be laid in the foundation for building a management decision support system.