Hybrid Modeling of Financial Ratios and Profitability: A Panel Regression and Random Forest Approach on LQ45 Firms in Indonesia

This study examines how four essential financial ratios—Current Ratio (CR), Debt to Asset Ratio (DAR), Net Profit Margin (NPM), and Total Asset Turnover (TATO)—affect Return on Assets (ROA) for Indonesia Stock Exchange LQ45 index firms from 2020 to 2023. The hybrid approach combines fixed-effect panel regression with the Random Forest algorithm to analyze 80 firm-year observations from 20 companies that met inclusion conditions for four years. The fixed-effect regression model shows that NPM, TATO, and DAR positively impact ROA, whereas CR is statistically negligible (β = 0.1791, p = 0.0525). The model explains 96% of ROA variation (Adjusted R² = 0.96). The Random Forest model has a predicted R² of 0.8246 and an RMSE of 0.0266, highlighting the predictive power of TATO and NPM. Although insignificant in regression, CR is moderately significant in Random Forest, but DAR's predictive power is minimal. Profitability and asset efficiency drive firm performance, but liquidity and leverage are more complicated and situational. Business managers should focus on margins and asset productivity, and regulators and investors should use econometric and machine learning insights to evaluate performance. This study strengthens the case for combining classical and machine learning models in financial analysis in emerging markets.