Many educators have worried about the failures of students through academic education. Thus, a variety of predictions have been applied to general information including culture, social, and economic information which wasn’t related to student performance. We have gathered an actual dataset from three years of academic stages of Mustansiriyah University in Iraq. The dataset consists of academic information without any socioeconomic data, it includes forty-four undergraduate students with thirteen attributes. We have proposed a model that explains the correlation between two main subjects which are, mathematics, and control systems. This study aimed to identify student failure of the control systems subject in the third year depending on the academic features of the mathematics subjects in the first and second years. Three algorithms were applied to the dataset including Naïve Bayes, support vector machine, and multilayer perceptron. Since the dataset was imbalanced, this leads to appear overfitting problem in the results so the synthetic minority oversampling technique was utilized to solve this problem. Our results show that the support vector machine algorithm proves an efficient classification after applied synthetic minority oversampling technique. The accuracy of the classifiers was measured from the confusion matrix using the Waikato environment for knowledge analysis (WEKA) tool and its related metrics.

In the multiple instances learning framework, instances are arranged into bags, each bag contains several instances, the labels of each instance are not available but the label is available for each bag. Whilst in a single instance learning each instance is connected with the label that contains a single feature vector. This paper examines the distinction between these paradigms to see if it is appropriate, to cast the problem within a multiple instance framework. In single-instance learning, two datasets are applied (students’ dataset and iris dataset) using Naïve Bayes Classifier (NBC), Multilayer perceptron (MLP), Support Vector Machine (SVM), and Sequential Minimal Optimization (SMO), while SimpleMI, MIWrapper, and MIBoost in multiple instances learning. Leave One Out Cross-Validation (LOOCV), five and ten folds Cross-Validation techniques (5-CV, 10-CV) are implemented to evaluate the classification results. A comparison of the result of these techniques is made, several algorithms are found to be more effective for classification in the multiple instances learning. The suitable algorithms for the students' dataset are MIBoost with MLP for LOOCV with an accuracy of 75%, whereas SimpleMI with SMO for the iris dataset is the suitable algorithm for 10-CV with an accuracy of 99.33%.

Classification under supervision is the most common job that performed by machine learning. However, most Educators were worried about the rising evidence of student academic failures in university education. So, this study presents a supervised classification strategy of machine learning algorithm using an actual dataset contains 44 students, fourteen attributes for three previous academic years. We have proposed features that show the relationship among three main subjects which are, calculus, mathematical analysis, and control system in the education course. The objective of this study is to identify the student’s failure in the control system subject and to enhance his performance by Multilayer Perceptron (MLP) algorithm. The dataset is unbalanced, which causes overfitting of the results. Synthetic Minority Oversampling Technique has applied to a dataset for obtaining balance dataset using Weka tool. Several standard metrics used to evaluate the classifier results. Therefore, the suitable results occurred after applying SMOTE with an accuracy of 76.9%.