loading page

Evolution and Optimization in Machine Learning: A Bibliometric Analysis and Strategy Overview
  • İpek DEVECİ KOCAKOÇ,
  • Meryem PULAT
İpek DEVECİ KOCAKOÇ
Dokuz Eylul Universitesi

Corresponding Author:[email protected]

Author Profile
Meryem PULAT
Dokuz Eylul Universitesi
Author Profile

Abstract

This study embarks on a comprehensive bibliometric analysis of 442 machine learning articles published between 1989 and 2019. Leveraging keywords such as “machine learning”, “hybrid machine learning”, and “hybrid decision tree” in the Google Scholar database, we collated articles and scrutinized them based on diverse parameters, including the language of publication, year, software employed, number of pages and authors, citation count, keywords, and methodologies utilized. Our research findings spotlight a noteworthy uptick in publications post-2008, predominantly in the English language. The most commonly used machine learning method, such as SVM, KNN, and ANN, have been determined. the most popular software tools like WEKA and MATLAB, have been determined. While feature selection methods, particularly the Genetic Algorithm and Information Gain, are widely used, there exists a marked underutilization of ensemble learning techniques. Within the scope of the study, the most frequently used parameter optimization methods were also determined. This study, which includes important information such as commonly used machine learning techniques, feature selection approaches and parameter optimization strategies, can guide researchers both as a retrospective overview of the development of machine learning and as a guide to increasingly used methods such as ensemble learning.