A data analytics framework for key financial factors

Purpose The purpose of this paper is to identify a subset of key financial ratios and factors that provide the best discriminating power to distinguish between creditworthy companies (CWCs) and less creditworthy companies (LCWCs) in the USA with the proposed method. Design/methodology/approach A proposed framework of Bisection Method Based on Tabu Search + Support Vector Machines (BMTS + SVM) is used to select subset of financial ratios from a pool of candidate ratios. The selected ratios and their corresponding financial factors are considered as the key financial ratios and factors that provide the best discriminating power to distinguish between CWCs and LCWCs. The authors collected financial data for the US companies and then identify the key financial ratios and factors which the selected key financial ratios belong. Findings It is found that the four selected financial ratios from the proposed method and eight financial ratios which are used by Standard & Poor for their credit-rating system can be attributed to the same four financial factors, namely, cash flow factor, profitability factor, solvency factor and leverage factor. This result lends support that the proposed method can be applied to identify key financial factors to differentiate CWCs and LCWCs. Practical implications This study provides a tool for managers in financial institutions to gain better understanding about the credit risk of their applicants by focusing on a parsimonious model with fewer ratios in the key financial factors. In addition, companies that attempt to borrow money from financial institutions can also use these key financial ratios and factors as reference to attain clearer vision on what are the most important factors for being considered a creditworthy company and thus develop specific strategies to improve their financial performance. Originality/value Based on data analytic techniques, this paper identifies key financial ratios and factors for examining the creditworthiness of US companies with the proposed framework using BMTS + SVM method.