# Predicting Bankruptcy in Wholesale, Retail, and Motor Vehicle Repair: An AI-ML Perspective

## TL;DR

This paper explores how AI and machine learning can predict bankruptcy in India's trade services sector, finding that tailored models work better than general ones.

## Contribution

The study introduces a segmented analytical approach for bankruptcy prediction in India's trade services sector using AI-ML models.

## Key findings

- AI-ML models accurately predict bankruptcy in Indian trade services firms.
- Segmented analysis reveals hidden risks not apparent in aggregate data.
- A tailored approach improves predictive accuracy and uncovers segment-specific risk factors.

## Abstract

Bankruptcy prediction is crucial for financial stability, and sector-specific Artificial Intelligence and Machine Learning (AI-ML) models have proven superior in performance. However, a significant gap exists, as most models are designed for advanced economies, leaving their efficacy in emerging markets like India unexplored. This study addresses this gap by focusing on the applicability of these advanced models to predict bankruptcy within India’s dynamic trade services sector.

The research utilized a substantial sample of 5,527 Indian companies. To counter the challenge of having far fewer bankrupt firms than solvent ones, the Synthetic Minority Oversampling Technique (SMOTE) was employed. The study then leveraged a comprehensive suite of eight popular AI-ML models, including Random Forests, Gradient Boosting, Neural Networks, and Support Vector Machines. To add practical context, business rules based on key financial metrics—liquidity, profitability, and asset size—were integrated.

The findings robustly demonstrate that AI-ML models can accurately predict bankruptcy in Indian trade services firms. A critical discovery was the variation in early warning signals between an analysis of the entire dataset (aggregate) and segmented groups of companies. This indicates that a one-size-fits-all approach obscures important, segment-specific risk factors. The segmented analysis successfully uncovered hidden risks that were not apparent at the aggregate level.

The study concludes that AI-ML models are highly effective for bankruptcy prediction in India’s trade services sector. For stakeholders like investors and creditors, the key takeaway is the superior value of a segmented analytical approach. This strategy maintains high predictive accuracy while revealing nuanced, specific risks. Ultimately, it provides a powerful, tailored tool for safeguarding financial interests in an emerging market context.