Analysis of Trust in Ukrainian banks based on Machine Learning Algorithms

Abstract

Purpose: This paper aims to conduct a thorough analysis to determine the influence of several possible factors on the level of confidence in Ukrainian banks. This scientific research project is based on data from the World Values Survey (WVS). The paper also aims to empirically investigate a number of independent variables, that creates trust in banks, based on Machine Learning Algorithms. Besides that, use some of the predictive analytics techniques to anticipate the level of trust in Ukrainian banks. Methodology: to calculate the index of Confidence in Banks var., by using linear regression, logistic regression (as a robustness check), Random Forrest, Decision Tree and XGBoost Models. Verify the output and results by using the Residual graphs, Cross-validation, Confusion matrix, ROC and model accuracy estimations. Main Findings: age, level of financial satisfaction, scale income and life satisfaction, general trust, lack of cash and other indicators has a significant impact on the the level of trust in Ukrainian banks. This paper represents a number and probability value of the variables spectrum; effect plots and other visualization graphs.