Improving classification performance for an imbalanced educational dataset example using SMOTE

Abstract

With technology, a lot of data is formed in digital environments. One of the areas with intensive data is educational data sets. By analyzing educational data sets, students' situatiokjgjjööÖns can be predicted by foreseeing. In this way, students can be assisted by anticipating situations such as drop-out due to failure. Educational institutions can take measures to prevent such dropouts and reduce student drop-out. Thus, financial losses of students and educational institutions can be prevented. In this study, the data of five separate associate degree students who were enrolled in Amasya University Distance Education Center in 2016-2017 were used. These are associate degree programs in child development, medical documentation and secretarial, electricity, mechatronics, and internet and network technologies. It was estimated whether the students could graduate or not at the end of the IV. Semester with looking at their I. and II. semester course notes. These data were analyzed by k nearest neighbor (K-NN) and KStar algorithms. Some of the data were obtained from the distance education center as imbalanced data due to the low number of students. In Educational Data Mining, researchers usually overlook the balance of the distribution on a dataset. Unbalanced data can seriously affect the success of classification. Synthetic minority oversampling technique (SMOTE) method was applied to these unbalanced data and how it affected the success of classification was examined. First, the raw data were analyzed with K-nearest neighbors classifier and KStar classifier. In this study, the analysis results of these five chapters are given in tables and comparatively. In this study, it has been seen that SMOTE oversampling method increase the classification success. In areas where unstable data such as educational data mining may exist, higher classification accuracy can be achieved with the help of different oversampling methods.

Ensemble-SMOTE: Mitigating Class Imbalance in Graduate on Time Detection

“…To develop and evaluate predictive models in identifying the likelihood of students graduating on time during their studies in university.   [11]  [12]    [13]  [14]  [15]  [16]  [17]  [18]  [19]  [20]  [21]  [22]  [23]  [24]  [25]  [26]    [27]  [28]  [29]  [30]  [31]  Table 1 illustrates the prevalence of minority classes within datasets utilized by researchers to tackle the challenge of class imbalance in identifying GOT. This imbalance arises when the dominance of the majority class eclipses the presence of minority classes, resulting in biased predictive models that yield unpredictable outcomes [12,15,16,27,28,30].…”

Handling the imbalanced data with missing value elimination SMOTE in the classification of the relevance education background with graduates employment

“…The most number of data in attribute the relevance educational background with graduates employment is owned by the class "not tight" namely 787 data, while there is number of the class less than 50% is "tightest" as much as 214 data, and "very tight" class as much as 304 data. The imbalanced data in each class can affect the accuracy, precision and sensitivity [10], [11]. The accuracy is the level of the success of the correct classification of all classifications for whole data regardless of the class of the data [12], so the accuracy is not enough for imbalanced data because the accuracy does not take the accuracy on each class of labels in output attribute.…”

“…The synthetic minority oversampling technique (SMOTE) is a simple method to get over the imbalanced data [10]. Several studies have shown that the SMOTE method can improve the accuracy of classifications as mentioned in studies [11], [16]- [18]. The imbalanced data is done at the pre-processing stage.…”

“…In Table 7, the accuracy of data in [17], [18], [27]- [28] was more than 90% but the increase of accuracy was less than other studies. Combination of SMOTE with hybrid scheme [18] increase accuracy than original SMOTE that applied in [11], [17], [27] and SMOTE that combine with axiomatic fuzzy SMOTE [28].…”