An Efficient Approach for Predicting the state of Financial Inclusion by Addressing Class Imbalance using the Oversampling Technique

The development of any nation can be improved by addressing financial inclusion issues. In many nations, there are challenges in getting people financially included in the banking system. As an example, Africa faces enormous challenges that affect development and livelihood due to poor financial inclusion reported in the Zindi financial inclusion dataset. The literature revealed that various machine learning techniques have been utilized to predict the state of financial inclusion but the performance of those approaches needs to be improved. In this paper, we have proposed an efficient approach for predicting the state of financial inclusion by addressing class imbalance issue in the Zindi dataset. So, to handle the issue oversampling has been done using Synthetic Minority Oversampling Technique. Further, we have utilized RandomForest, XGBoost, DecisionTree, Logistic Regression, Support Vector Machine K-Nearest Neighbors, GradientBoosting, HistogramGradient Boosting, and AdaptiveBoosting for predicting how likely a person will have a bank account. The holdout cross-validation method has been used to split the training dataset randomly into 90% train and 10% test set. The result was collected and further compared with some existing approaches, which indicates that our approach outperforms the existing state-of-the-art approach. The Random Forest was optimal among adapted models based on F1-score, precision, recall, and accuracy metrics which are 91.556%, 91.357%, 91.736%, and 91.541% respectively.