In order to determine the gap and novelty in credit score research, the study begins by looking at the quantity and distribution of research articles pertaining to credit scores, pay later, and payment behavior keywords over the specified study period. Next, it undertakes a bibliometric analysis of credit score measurements found through a literature search until August 2022. Lastly, it outlines credit score research subjects based on keywords and author. In conjunction with categories, article titles, abstracts, and the word "key," data is gathered from Scopus between August 2022 and the year 2017 using the terms "credit scores," "pay later," and "payment behavior." The trend of publication developments was determined by examining data in the form of the number of publications per year, journals containing keyword articles, authors, and origins of authors using the software applications VOSviewer and RStudio. The findings revealed a yearly increase in the quantity of publications pertaining to this research, culminating in a total of 317. Machine learning is the predominant technique utilized in the assessment of credit scores. The United States is the largest contributor to the ACM International Conference Proceedings series and Advances in Intelligent Systems and Computation, which are the most prolific sources of research on this subject. Zhang Y is a prolific writer. By conducting gap analysis and novelty research using RStudio and VOSviewer, it was ascertained that while studies have been conducted on credit card analysis, peer-to-peer lending, and loan banking systems utilizing machine learning to calculate credit scores, there is a dearth of research on pay later services. Moreover, numerous credit score categorizations are predicated on exemplary and inadequate credit performance. Consequently, the development of supplementary metrics for credit score levels or ratings becomes feasible.