|
Article information
2022 , Volume 27, ¹ 1, p.52-69
Blinov P.Y.
The methodology for classifying time series by trend type in scientometrics
Nowdays reasearchers often encounter the problems dealing with processing large amounts of data and so there is a growing interest in various quantitative metrics in different areas. In particular, a number of metriñs characterizes publication activity in scientometrics. Many countries are currently experiencing an increase in publication activity, moreover, some countries have a tendency of abnormal growth for some reasons. The methods of mathematical statistics are quite effective for identification of certain growth patterns. In this paper, usefulness of statistical tests of randomness or trend tests for time series data in scientometrics is demonstrated. Methodology of aggregate testing by several trend tests is proposed. This methodology, being a very flexible tool, is focused on improving the accuracy of estimation for small size time series as well as allows a large number of time series to be checked and grouped by type of trend.
[full text]
Keywords: scientometrics, publication activity, statistical hypothesis testing, randomness test, power of test, trends, significance level
Author(s):
Blinov Pavel Yurievich
PhD.
Position: Senior Research Scientist
Office: Russian Research Institute of Economics, Politics and Law in Science and Technology
Address: 127254, Russia, Moscow, 20A, Dobrolubova Str.
E-mail: p.blinov@riep.ru
SPIN-code: 4065-4149
Bibliography link:
Blinov P.Y. The methodology for classifying time series by trend type in scientometrics // Computational technologies. 2022. V. 27. ¹ 1. P. 52-69
|
