Identify the Economic Crisis by Analyzing Banking Data Using Machine Learning Technique

Abstract

The aim of this research is to pinpoint economic downturns by delving deep into banking data and scrutinizing the countermeasures banks employ to avert these downturns. By amalgamating quantitative assessment of banking metrics with qualitative insights from bank documentation and dialogues with banking leaders, we offer a comprehensive perspective. Key indicators that signal economic crises include escalating numbers of non-performing loans, waning profitability, and shrinking capital ratios. In response, banks have adopted strategies such as reshaping their loan portfolios, bolstering capital reserves, and refining their risk assessment protocols. This research underscores the paramountcy of promptly detecting economic downturns and deploying efficacious strategies to safeguard financial equilibrium. We structured our approach in three pivotal phases: firstly, countering overfitting through regularization; next, utilizing boosting to minimize loss during each iteration; and finally, addressing edge cases. Our proposed framework commences with data partitioning. This is succeeded by statistical preprocessing, yielding a detailed multivariate analysis for the ensuing model. This refined data then feeds into our boosting model, which, after training, facilitates classification. The outcomes of this classification stage are then channeled into a regression model, delineating the ramifications of exchange rate fluctuations due to economic upheavals.