Machine Learning With Ensemble Classifier Algorithm For Imbalanced Data Classification

Abstract

Almost all fields of real-world study have difficulties while performing data analytics because of data imbalances. Data on security or marketing or healthcare suffer from skewed class distributions. To deal with the categorization of unbalanced data, ML (Machine Learning) and EC (Ensemble Classifier) techniques are suggested in this study. Main stages of this job include pre-processing, class balancing, and classification. The datasets are first gathered, and the KMC (K-Means Clustering) method is used to do pre-processing. By filling in the blanks, the categorization accuracy is being increased. The SMOTE-LOF approach is then used to include the datasets into the class balancing procedure. It conducts under- and oversampling as well as outlier identification. In order to improve classification accuracy, a basic classifier with the EC algorithm is used next. Base classifiers in this study include ensemble methods like bagging, boosting, and stacking models, as well as ML (machine learning) algorithms like EANN (Enhanced Artificial Neural Network), KNN (K-Nearest Neighbour), and SVM (Support Vector Machine). Better classification outcomes are produced as a result of the EC algorithm's higher base classifier accuracy. The performance of the balanced dataset is significantly improved by the suggested EC model, according to experimental findings. The results showed that, compared to the current techniques, the suggested base classifier with EC algorithm offers improved accuracy, precision, recall, F-measure, AUC (Area Under Curve), and reduced execution times.