Classification of Imbalanced Data Represented as Binary Features

Abstract

Typically, classification is conducted on a dataset that consists of numerical features and target classes. For instance, a grayscale image, which is usually represented as a matrix of integers varying from 0 to 255, enables one to apply various classification algorithms to image classification tasks. However, datasets represented as binary features cannot use many standard machine learning algorithms optimally, yet their amount is not negligible. On the other hand, oversampling algorithms such as synthetic minority oversampling technique (SMOTE) and its variants are often used if the dataset for classification is imbalanced. However, since SMOTE and its variants synthesize new minority samples based on the original samples, the diversity of the samples synthesized from binary features is highly limited due to the poor representation of original features. To solve this problem, a preprocessing approach is studied. By converting binary features into numerical ones using feature extraction methods, succeeding oversampling methods can fully display their potential in improving the classifiers’ performances. Through comprehensive experiments using benchmark datasets and real medical datasets, it was observed that a converted dataset consisting of numerical features is better for oversampling methods (maximum improvements of accuracy and F1-score were 35.11% and 42.17%, respectively). In addition, it is confirmed that feature extraction and oversampling synergistically contribute to the improvement of classification performance.

Predictive Data Modeling: Student's Obstacle in Mathematical Literacy Tasks Focusing on Ratio and Proportion using The K-Nearest Neighbor Algorithm

“…Accuracy provides information about the extent to which the model predicts classes correctly. Balanced accuracy becomes relevant if the dataset has a class imbalance, whereas the F1 Score delivers a balance between precision and recall, especially in the case of binary classification (Mahmudah et al, 2021).…”

Performance analysis of samplers and calibrators with various classifiers for asymmetric hydrological data

“…When dealing with asymmetric data classification, different evaluation metrics were essential. While accuracy is a common metric for classification [6062], it is not suitable for asymmetric classification [63] because a less effective model can achieve a higher accuracy. Evaluation of expected and predicted class labels or assessment of probabilities for the anticipated class labels were needed for classification issues.…”

“…Additionally, the AUC analysis was employed to visualize the categorization of the dataset used for label prediction. Equations 1 to 5 provide the mathematical expressions for calculating specificity, sensitivity, BRA, G-mean, and AUC, as follows: Specificity: The percentage of accurately detected negatives over all possible negative forecasts produced by the algorithm is measured by specificity [63]. It is also referred to as the true negative rate.…”

“…Sensitivity: A performance metric derived from the positive observations is sensitivity or recall [63]. The percentage of positive observations that were identified accurately as positive is displayed by sensitivity.…”

KNN Based Under Sampling: A Cognitive Centred Solution for Imbalanced Dataset Problem in Anaphora Resolution

“…Table 1 presents the summary of prune rules derived for the present scenario. https://www.indjst.org/ List of pruning rules is applied over the feature vector to reduce the difference between positive and negative pairs before it is given to classifiers (11) . During experimentation number of positive instances is increased from 5.32% to 43.95%, whereas the number of negative instances is decreased from 94.68% to 56.05%.…”