Introduction

The rapid development of digital transformation and artificial intelligence technologies is one of the main issues that ensure the world economy enters a new stage of development. Technological development has led to a significant change in global economic relations and the emergence of a new generation of technological economic sectors. The next-generation economic sectors are characterized by increased integration of communication, automation, artificial intelligence, and other digital technologies and innovations. In the current period, the application of Machine learning methods, one of the artificial intelligence technologies, is very relevant in improving the processes of formation and development of the traditional sectors of the economy, and in forecasting its new generation perspective sectors. Machine learning methods and artificial intelligence technologies are considered to be the main methods that allow more accurate and effective forecasting of future economic areas [5, 8].

The role of machine learning in the accurate analysis of economic processes and forecasting the development of prospective economic sectors is indispensable. In addition to traditional economic activities, the new generation economy envisages the development of a renewed economy consisting of digital finance, biotechnology, green technology, green energy, and sustainable technologies. In order to conduct an effective economic policy and make the right economic decisions in these areas, it is required to make highly accurate and long-term forecasts [8, 24].

It is no coincidence that in the Socio-Economic Development Strategy of Azerbaijan technological innovation factors include 1) effective integration of the country's economy into the global value chain, 2) through technological innovations, as well as state strengthening of resource and service provision in this field with private partnership, 3) formation of the necessary ecosystem for rapid adoption of technological innovations in the national economy, 4) transformation of the country into a green energy space, and 5) acceleration of environmental health processes, etc.

Statement of the problem

Through the application of machine learning in the development of the economy, the analysis of information about new sectors of the economy is accelerated and areas with development potential are identified more quickly. At the same time, it becomes possible to predict future market demands, determine the development of various industries, and create more stable decision-making systems [17].

The formation of new generation technological economy sectors and technological development is reflected in many State Programs and official documents, where technological development is one of the main tasks for advanced countries, including Azerbaijan. In those programs, taking into account the great importance of technological development in the field of industrial development and the civil and military-industrial complex, its main goals were defined. It has been shown that technological development, digitalization, cyber security, and the application of artificial intelligence have become signs of daily life in the country and have been set as a task before the institutions and society.

Analysis of scientific research related to the problem

It is known that the application of cognitive technologies in solving many problems of digital innovative technologies in the economy, forecasting some of its processes and effective management has created new opportunities and conditions. Certain scientific studies have been conducted on this issue. It should be noted that the authors of the articles, like other researchers, have publications in this field in different years and in scientific journals indexed in different high scientific databases [1-5, 7, 8, 24]. The requirements for solving the relevant problems in those publications must be taken into account in the comprehensive formation of the new generation technological economy and in the analysis of development problems. For this reason, despite the fact that the analysis of the above-mentioned problems has been studied to a certain extent in the scientific works of many scientists and specialists, the demand for a deeper and comprehensive study of that process, as well as the development of various approaches to solving the problems that have arisen, still remains and remains relevant.

In [21], the issues of macrostructural analysis and forecasting of economic development in modern conditions were considered. Opportunities to assess the medium and long-term potential of economic growth using macroeconomic and intersectoral tools were presented. [23] is dedicated to a new paradigm for stimulating and forecasting China's economic growth in the medium and long term. Here, the system philosophy of certain relations is considered theoretically. Here, the supporting system of China's economic development is established for the three dimensions of economic growth, social development, and environmental protection. Based on this, based on the principles of system theory and system dynamics, and as a result of combining the theories of other related disciplines, an economic geography-system dynamics integrated forecasting model was built to simulate and quantitatively forecast China's medium and long-term economic growth.

In [16], the development of the Deep LSTM method for predicting the price of Bitcoin, one of the cryptocurrencies of the new generation economy, was considered. Here, a deep neural network architecture is proposed to predict the exchange rate of Bitcoin. The architecture is based on LSTM (Long-Short Term Memory), a type of recurrent neural network. The obtained results allow us to confirm that deep learning approaches can be quite productive in applying them to other issues of intellectual analysis of non-stationary time series related to cryptocurrencies.

A bibliometric analysis of published literature on machine learning in economics and econometrics was conducted in [12]. The purpose of the study is the analysis of data collected from publications indexed by the Web of Science and Scopus databases covering the years 1991-2020. Analysis of variance was used to determine the relationship between the number of article citations and years. In [15], Internet financial forecasting and digital economy development issues were analyzed using machine learning algorithms in the new consumer environment.

In [8], the conceptual bases of the application of artificial intelligence technologies in the forecasting of economic processes were developed. Artificial intelligence methods based on evolutionary algorithms, Data-mining technology, machine learning technology, and computer vision were analyzed in forecasting economic processes. The basics of applying the neoclassical production function in predicting the impact of artificial intelligence on economic growth have been investigated. Neoclassical models were used to investigate the impact of artificial intelligence technology on production. Based on data mining technology, the economic development potential forecasting model and its forecasting adequacy were explained. The structural scheme of the main components of the application of artificial intelligence technology in the forecasting of economic processes has been developed.

In [14], issues of cooperation between the state and business in forecasting and planning structural changes in the economy were considered. In this work, the Japanese experience of indicative planning during the reconstruction of the economic structure is presented. Based on a long-term view of the Japanese economy, the degree of involvement in the business planning process and its evaluation have been noted. In [22], the issue of predicting the income of the management company based on machine learning technologies was analyzed. Here, an algorithm based on the machine learning method is proposed for forecasting the earnings of a management company.

[13] on inflation rate forecasting and machine learning method analysis. Here, the analysis covered the years 2012-2022. It has been shown that the Random Forest model can predict the inflation rate more accurately than other machine learning models. Also, non-linear machine learning models have been proven to be more successful than linear machine learning or time series models.

New directions for applied knowledge-based artificial intelligence and machine learning are reviewed in [11]. Selected new directions in knowledge-based artificial intelligence and machine learning are presented here: ontology development methodologies and tools, automated engineering of WordNets, innovations in semantic search, automated machine learning, and AutoML. Knowledge-based AI and machine learning complement each other ideally, as their strengths compensate for each other's weaknesses. These are demonstrated through selected enterprise use cases (anomaly detection, efficient modeling of supply networks, circular economy, and semantic enrichment of technical information).

In [18] inflation forecasting in developing countries, machine learning for advanced forecasting and the integration of foreign exchange reserves are considered. Research results have shown that machine learning models consistently outperform traditional models, with Random Forest and Gradient Boosting being the best across different sets of determinants. In addition, the study finds that the inclusion of foreign exchange reserves as a determinant in the models has a positive effect on the predictive performance of both traditional and machine learning-based inflation forecasting models. As can be seen from the analysis, many aspects of the application of the machine learning method should be taken into account in the complex forecasting of the formation of prospective sectors of the new generation economy.

Applications of new generation digital technologies and transformation in the economy

Cognitive technologies, which constitute the intellectual foundations of new generation digital technologies [6] are essentially software and equipment that imitate the activity of the human brain and work with the user. They try to assess a person's attention, monitor his condition, help his brain function, and generally "understand" him. These technologies also include artificial intelligence and machine learning. 1)Artificial intelligence and machine learning are revolutionizing the activities of various economic sectors. These technologies allow analyzing large volumes of data, improving decision-making, and automating complex tasks in the economy and business. Blockchain technology provides a decentralized, secure, and transparent way to record transactions. Its applications cover all industries, contributing to the development of new business models and digital ecosystems. Blockchain technology is used in trade financing, P2P transactions, and smart contracts, by reliably protecting the history of all recorded transactions and processes without falsification, based on a distributed (decentralized) database. During the emission ("production", mining) of cryptocurrency as a virtual currency, it is based on the special application of cryptographic algorithms. Internet of Things (IoT). technology also refers to an interconnected network of physical devices, sensors, and software that collect and exchange data. IoT has numerous applications in sectors such as agriculture, manufacturing, healthcare, and smart cities. The Internet of Things (IoT) is a set of technologies that involves equipping various devices and equipment with sensors and connecting to the Internet for remote monitoring and control of processes in real-time (including automatic mode). The Internet of Things, in particular, allows tracking the movement of goods, providing remote service, and supporting customer self-service.

The basic principles of the formation of the new generation technological economy and the problems of its formation

The next generation economy is fundamentally reshaping the way value is created, distributed, and stored globally. Its formation is governed by a unique set of principles that reflect changing technological, social, and economic paradigms. These principles form the basis of a more flexible, innovative, and sustainable economic system adapted to the complexities of the 21st century. The rapid evolution of digital technology, and its integration into every area of society, has led to the emergence of what is now called the next-generation economy. This economic paradigm represents a transition from traditional models to new foundations driven by digital transformation, global connectivity, and sustainability [9, 10].

Features and prospects of application of machine learning methods in forecasting prospective sectors of the new generation economy

Since machine learning is a multidisciplinary field, it interacts with mathematical statistics, computer science, applied mathematics, psychology, physiology, etc. It is an interdisciplinary course that combines knowledge of probability theory, estimation theory, mathematics, and statistics. These methods are based on sophisticated algorithms that use computers as tools designed to mimic how humans learn in real time [2, 17]. The content of machine learning 1) Machine learning is an implementation of artificial intelligence that improves the efficiency of empirical learning algorithms. 2) Machine learning is the study of computer algorithms that can be automatically improved through experience. 3) Machine learning uses past data to improve the performance of computer programs. Data collection, data operations, data analysis, training, and testing stages are performed to solve problems with machine learning methods. Machine learning in Computer Vision, Speech Recognition, Image Recognition, Computer Linguistics and Natural Language Processing, Medical Diagnostics, Technical Diagnostics, Information Retrieval, etc. applies.

Machine learning is divided into several main categories: 1) Supervised learning. The model is presented with the correct answers beforehand. The system learns to respond correctly to new information by learning from these patterns. 2) Unsupervised learning. The model is given unanswered and uncertain data. The system tries to discover hidden patterns and structures among that data. 3) Reinforcement learning (Reinforcement Learning). The model learns based on experience. He receives grades in the form of rewards and punishments for certain decisions or actions. As a result, he learns to make good/correct decisions over time. This technology creates new opportunities in the world of science and business. Thus, it allows to analyze the data more accurately and make smart decisions using them. The classification features of machine learning methods can be expressed as in Figure 2.

Machine learning models can be attributed to: The choice of the machine learning models as Logistic Regression (LR), K-Nearest Neighbor (KNN), Naive Bayes (NB), Decision Tree (DT), Support Vector Machine (SVM), and Artificial Neural Network (ANN), and ensemble methods as Random Forest (RF), and Adaptive Boosting (AdaBoost) are based on the most commonly used models in the reviewed works [20].

Decision trees (DT) method is used to predict discrete-valued objective functions. Here, a decision tree represents a learned function that combines predictor variables into an expected variable [19]. To search for the most discriminating variables, the decision tree algorithm uses a corresponding approach to construct a tree-like structure of nodes and edges. [19] used heuristic measures such as the Gini index, entropy, etc. in a research study on how the decision tree model works to determine the most discriminating variable values. Logistic regression is a nonlinear regression technique that associates a conditional probability score with each data sample. The essence of logistic regression is to examine the relationship between dependent and independent variables. The dependent variable can be binomial or multinomial.

About stages and methods of forecasting economic development

Machine learning methods applied in forecasting. A comparative analysis of machine learning methods applied in forecasting shows that when choosing them, some of the effective machine learning methods designed for the prediction of large amounts of data can be identified as in a separate table [22]. Their advantages and disadvantages are presented as in table 1.

Stages of application of the machine learning method in the forecasting process

Machine learning is an artificial intelligence approach that allows models to make predictions based on previous observations [11]. Machine learning is used to power AI applications (figure 4). Building effective AI applications based on machine learning is complex. The steps to be performed include data preparation, characterization, model selection, parameter optimization, validation, etc.

Conclusion