Scientific journal NRU ITMO
Series "Economics and Environmental Management"
Registration certificate ЭЛ № ФС 77 – 55411 от 17.09.2013
registered by the Federal Inspectorate Service for Communication, Information Technologies and Communication Media
ISSN:2310-1172

June 2020 (published: 19.06.2020)

Number 2(41)

Home > Issue > Using machine learning in investment activity

UDC 330.43

Using machine learning in investment activity

Kazanskaya A.A. , Mishura L.G.

We probably live in the most defining time for information technology. The period when computing switched from large electronic computers to automatic control machines and robots, with the ability to store data in the cloud. But what makes this period truly exciting, is the democratization of various tools and methods that developed simultaneously with the development of computer technology. Data processing, which once took several days, today takes only a few minutes, and all this thanks to the development of machine learning. This is the reason why Data Science receives huge investments every year, which increases the demand for certificates in this area. Russian venture capital company and partner of the investment company iTech Capital Alexey Soloviev presented the study “Venture Barometer” on the Russian venture market. 312 representatives of the venture capital market were invited to participate in this study, of which 83 answered the questionnaire. The barometer annually asks investors about the most attractive market segments, and in 2019, as in the past 2 years, the ranking is headed by artificial intelligence and machine learning. 82% of investors voted for this field as the most promising [15]. Data and Methods: The objective of this study is to build an optimal algorithm that predicts the rating value of an investment project for a given data set. The most popular machine learning algorithms are examined in detail, a detailed analysis of the available data is carried out, a model based on a random forest algorithm for a training data set is constructed, and testing is conducted based on test data. This research work was performed in the programming language R, with the help of which the data were analyzed, various tables and graphs were constructed, a model was built and the results evaluated based on a comparison with previously constructed algorithms: linear regression, logistic regression. Analysis of Results: The study revealed that the most optimal algorithm for predicting the rating of an investment project based on available data is the random forest algorithm, the accuracy of this algorithm is 3.7% higher than the accuracy of the linear regression algorithm based on the most significant set of project indicators.
Read the full article 

Keywords: Information technology, investment activities, innovation, machine learning, random forest algorithm

DOI 10.17586/2310-1172-2020-13-2-23-34

Editorial address:
191002, St. Petersburg, street. Lomonosova 9, office. 2132

Information 2007-2020, All rights reserved
Department of Internet Solutions NRU ITMO © 2013
Яндекс.Метрика