Modified Grey Wolf Optimizer with Sparse Autoencoder for Financial Crisis Prediction in Small Marginal Firms

• Published in: 2023 Second International Conference on Electronics and Renewable Systems (ICEARS) pp. 907 - 913
• Main Authors: Bhattacharya, Rajib, Kafila, Krishna, Somanchi Hari, Haralayya, Bhadrappa, Nagpal, Pooja, Chitsimran
• Format: Conference Proceeding
• Language: English
• Published: IEEE 02.03.2023
• Subjects: Data preprocessing; Financial crisis; Grey wolf optimizer; Machine learning; Prediction algorithms; Predictive models; Renewable energy sources; Small marginal firms; Training; Transforms
• DOI: 10.1109/ICEARS56392.2023.10085618

Small marginal firms play an important role in the economy, and their failure has widespread consequences. As a result, it is critical that small marginal firms be capable of predicting financial crises in order to mitigate their negative impact. Financial Crisis Prediction (FCP) is the process of recognizing the possibility of a future financial crisis. FCP is an important task for financial institutions, policymakers, and investors because it helps them to prepare for and mitigate the negative impact of financial crisis. Machine learning (ML) approaches are used to forecast financial crisis in small and marginal firms. It is completed by training a model on historical data and using it to predict the likelihood of a future financial crisis. As a result, this article proposes a Modified Grey Wolf Optimizer with Sparse Autoencoder (MGWO-SAE) for predicting financial crisis situations in small marginal firms. The goal of the MGWO-SAE technique is to forecast financial crisis effectively in small marginal firms. To accomplish this, the proposed MGWO-SAE technique employs data preprocessing to convert the input financial data into the correct format. The MGWO-SAE technique employs the SAE classification method for prediction. The MGWO algorithm is used for hyperparameter tuning model to improve the performance of the SAE technique. The MGWO-SAE technique's experimental result analysis is tested on a financial dataset. The experimental results demonstrated the advantages of the MGWO-SAE technique over other models.