Resampling Imbalanced Class and the Effectiveness of Feature Selection Methods for Heart Failure Dataset

Abstract

The real dataset has many shortcomings that pose challenges to machine learning. High dimensional and imbalanced class prevalence is two important challenges. Hence, the classification of data is negatively impacted by imbalanced data, and high dimensional could create suboptimal performance of the classifier. In this paper, we explore and analyse different feature selection methods for a clinical dataset that suffers from high dimensional and imbalance data. The aim of this paper is to investigate the effect of imbalanced data on selecting features by implementing the feature selection methods to select a subset of the original data and then resample the dataset. In addition, we resample the dataset to apply feature selection methods on a balanced class to compare the results with the original data. Random forest and J48 techniques were used to evaluate the efficacy of samples. The experiments confirm that resampling imbalanced class obtains a significant increase in classification performance, for both taxonomy methods Random forest and J48. Furthermore, the biggest measure affected by balanced data is specificity where it is sharply increased for all methods. What is more, the subsets selected from the balanced data just improve the performance for information gain, where it is played down for the performance of others.

A proposed method for handling an imbalance data in classification of blood type based on Myers-Briggs type indicator

“…The resample has improved classification performance significantly, even on highdimensional data without using the feature selection method in line with previous studies. This previous research has shown that the feature selection methods' role, except for the information gain, is still lesser than the resample method to improve the classification performance, including accuracy, sensitivity, and precision [20].…”

“…SMOTE can obscure the information of data interrelations in every class, thereby reducing the k-NN performance. As in previous studies, the information of interrelation data can be obscured due to the addition of unique samples that lead to generalization errors in the classifier [10], [18], [20]. Because of this flaw, SMOTE needs to be developed as in previous studies to improve the interclass boundary [10], [14], [23].…”

“…High-dimensional data is also a challenge in data mining because it can increase calculation complexity in data interpretation and potentially reduces classification performance [20]. Besides that, the data in this research is possibly unbalanced.…”

“…Oversampling replicates data from minority classes to match the majority and adds information between instances in minority classes. The undersampling process reduces the majority class's frequency by replacing or deleting some data in the sample, so the data's composition is balanced [20]. However, undersampling has a risk of losing data in the majority class if the data apparently may improve the classification process [23], [24].…”

Oversampling Method on Classifying Hypertension Using Naive Bayes, Decision Tree, and Artificial Neural Network

“…Kelas yang memiliki banyak instance disebut kelas mayoritas dan yang memiliki jumlah instance yang lebih sedikit disebut kelas minoritas [10]. Dalam situasi kehidupan nyata kadang-kadang kelas minoritas lebih menarik daripada kelas mayoritas, misal dalam data medis pada kasus gagal jantung [11] bidang ekonomi seperti credit scoring [12], credit cards fraud, dimana penyalahgunaan kartu kredit lebih sedikit daripada yang tidak disalah gunakan. serta bidang-bidang lain seperti deteksi spam pada email dimana email berupa spam lebih sedikit daripada bukan spam.…”