Forecasting of Bank Performance using Hybrid Machine Learning Techniques

Banks are the most powerful economic stimulants, and they play a crucial role in state economic development since they control the supply of currency in circulation to a substantial extent. To sustain a healthy financial system and an efficient economy, the forecasting of the bank can be considered as an important part of their financial planning process in the banking industry. So, to achieve this essential requirement of the banking sector, previously many attempts have been made where people have made forecasting systems using Binary Gradual Analysis, Data Envelopment Analysis (DEA), Linear Discriminant Analysis, optimizing machine learning approaches, specifically Particle Swarm Optimization (PSO) and Random Search (RS), etc. In our study, predictive analysis is carried out on 100 banks’ performance datasets using Support Vector Machine (SVM), Decision Tree (DT), and Artificial Neural Network (ANN) techniques. The dataset includes the value of 29 performance indicators that measure performance from 2000 to 2020. In addition, we attempted to create a few models by hybridizing SVM, ANN, and DT techniques such as SVM_ANN, SVM_DT, DT_ANN, and SVM_DT_ANN. Finally, a comparative analysis was done between hybridization models and their standard versions.