The Philosophy of Statistical Modeling

Abstract

This research is a literature review study aimed at exploring the essence of the philosophy and the underlying rationale behind statistical modeling. The research method began with the identification of relevant keywords, namely “Statistical Modeling,” “Why Modeling is Needed,” “Operational Model,” “Estimation,” and “Hypothesis Testing.” The literature search was conducted through credible sources such as international journal databases (Scopus, Web of Science, PubMed, IEEE Xplore), Google Scholar, university digital libraries, as well as reference books and conference proceedings. Based on these keywords, the initial search resulted in approximately 435 articles. After an initial screening based on the titles and abstracts in accordance with the inclusion and exclusion criteria, 75 relevant articles were obtained. A further selection process through full-text reading and quality evaluation of the literature resulted in 30 articles that were reviewed in depth, with the entire process documented using a PRISMA flow diagram and supported by reference management software such as Mendeley, Zotero, or EndNote. The research findings indicate that statistical modeling plays a central role in the modern data-dominated era. Various approaches, such as regression, Bayesian methods, and probabilistic models, allow for the simplification of data complexity, analysis of relationships among variables, and reliable predictions. The Bayesian approach, in particular, stands out due to its ability to handle uncertainty and causal relationships through graphical models such as Bayesian networks, which have proven effective in artificial intelligence and sequential data analysis. Furthermore, probabilistic models provide a representation of uncertainty through probability distributions, which is crucial in the analysis of complex data. The main advantage of statistical modeling lies in its functions in simplification, representation, and prediction, which are supported by the integration of advanced technologies such as machine learning. The process of parameter estimation and hypothesis testing ensures the validity and reliability of the model, making statistical modeling an essential tool in decision-making and practical applications across various fields.