Machine learning: a bibliometric analysis

Objective: Present an overview of scientific articles published in the last ten years on the topic of machine learning (ML), with an emphasis on predictive algorithms. Method/approach: Bibliometric analysis, with support from the PRISMA protocol, to evaluate authors, universities and countries, regarding productivity, bibliographic citations and focuses on the topic, with a sample of 773 articles from the Scopus and Web of Science databases, from 2013 to May/2023. Originality/value: There is an absence of studies in the literature that consolidate articles related to ML and Big Data. The research contributes to covering this gap, favoring the design of future actions and research. Main results: The following were identified in the ML bibliometric corpus: most cited authors with the greatest number of publications, most productive countries and universities, journals with the greatest number of publications and citations, areas of knowledge with the greatest number of publications, and the most prestigious articles. In the ML themes and domains, the following were identified: main co-occurrences of keywords, emerging themes (grouped into five clusters), and word clouds by title and abstract. Studies on the impact of data acquisition and predictive analysis represent opportunities for future research. Theoretical/methodological contributions: The PRISMA protocol enabled the identification and relevant quantitative and qualitative analyzes of articles, consolidating scientific knowledge on the topic. Social/managerial contributions: Ease of understanding the maturity of research on ML and Big Data by company managers and researchers, regarding the feasibility of investments to obtain competitive advantages with such technologies.