The performance of the banking industry is one of the main indicators of economic growth. It plays a vital role in various socioeconomic activities. A strong banking sector is essential for a robust economy. The poor performance of the banking sector in terms of financial risk management may adversely impact the other sectors of the economy. In India, non - performing asset (NPA) is a key factor that enhances the credit risk substantially for any bank. The performance of the public sector banks in risk management in the recent past years has been declining in view of NPAs. The ability of the banks to identify defaulters before lending is paramount for minimizing the incidence of NPAs as well as developing effective mechanism to proactively deal with potential defaulters. Various financial indicators such as quick ratio, profit after tax (PAT) as percentage of net worth, total net worth, and cash profit as percentage of total income will enable the concerned authority to spot possible defaulters and take appropriate corrective measures. With this background, an attempt was made in this paper to study key factors leading to non - performing assets. This research study focused on how the key factors impact NPAs based on insights derived from three important classifications and predictive models namely random forest (RF), gradient boosting machine (GBM), and logistic regression. The findings of this study will pave way for policy makers in banks to assess the probability of borrowers repaying the loan and classify them as good credit or bad credit.