Predicting Students’ Academic Performance: A Review for the Attribute Used

This article is a review on attributes, models and tools used to solve the problem of predicting students' academic performance. Based on the reference throughout 2009, the attributes used include demographic, academic grades, school related and social attributes. The academic score attributes are the results of the percipient along their studies, while the school-related attributes are result for few subject taken in high school. Demographic and social attributes are family background and the daily interactions of respondents. These tools are supporting a lot of rule mining algorithms like clustering, classification and association, to use on datasets of different types.


Introduction
The focus objectives of the study are predicting the students' academic performance, CGPA, grades, score and performance (not related to education). There are 28 articles that are related to this study. However, the articles collected different type of data set to carry out the prediction process. In this subtopic, these articles will be discussed more specifically.   and  has the same main objective. This objective can be achieved by using different types of data collected from data set or sample, based on different projects. There are few learning algorithm used as mentioned in Hachem & Alaou (2019); Anuraag, et al (2020); Yan & Abas (2020); Kangungu & Yatim (2020); Jianan Abas (2020); Maulana & Normalisa (2019); baba et al (2014); Desai & Singh(2020) ;Yusuf, et al, (2010);and Yusuf, et al, (2010).

Attribute Used
For article [1], the data set collected for the project is the grades got from the Computer Science undergraduate courses students of the chosen university in the research that got offered in the first, second and third semesters. Then, for article [2], data set used included the attributes that has information of preregistration students from a registration office. While for article [3], it took 61 students dataset of 2018 academic year that was obtained from the Department of Networking and System Security in KSITM. First Year CGPA Float 10 Predicted Class of Graduation

Nominal
It contains some students' information and academic data such as Grade Point Average of first semester (as GPA_1) which is numeric variable and number of subjects failed from preceding semester. Article [4] took participants consist of undergraduate students that had been studying Engineering or Psychology, to complete answering some of the long-term memory tests for the process of collecting the data needed. For the successful of project of article [5], it used attributes that consist of some instances that also included the student grades, social, demographic and school related attributes. As for article [6], to predict the students' academic performance, it used the data set that includes attributes of the students such as demographic, academic grades, school related and social attributes. Besides, to create algorithm for predicting students' academic performance in Tiwari, et.al (2019), only student's academic data needed with other existed data algorithm. Other than that, article by Baijam & Lenin (2019), a total of 79 observation were selected consist of 11 different attributes. While, only need engineering students' final grades for the prediction data set (Editha, 2019). Another article, by Ranjeeth, et.al. (2019), collecting data by asking 42 different questions as shown in Table 2.  (2019) where they need data consist of 27 variables that can be refer at Table 3.  (2019) predictive task had been carried out for students to collect the data. To predict student performance, we can also use students' academic information such as article in  that collect data from a total number of transcripts of the year 2013 to 2016 from the students who had completed their degrees in academic.
For article , it shows the review on several other articles and conclude that crucial factors to predict academic performance of students are personal attributes, family attributes, social attributes, student attributes, academic attributes, and school attributes.
Lastly, on article panessai, et. al. (2019) the researcher predicts the academic performance by using the data collected from the transcript of students from Universiti Pendidikan Sultan Idris that majoring in Software Engineering Program. This data is from the year of 2015.
The attributes for the data set is shown in Table 4. The next objective that going to be discussed is predicting the students' academic CGPA (Aderibigbe & Odunayo, 2019). This article obtained the data needed by collecting the first three academic years GPA data of some students and their final CGPA from the year of 2002 to 2014, from some engineering departments to achieve the objective.
For the next objective is predicting the students' grades based on articles Sara & George (2019) (2019), the grade prediction is carried out, by learning the weights of the prior courses towards predicting the grade of each target course. As for article Gamie et.al (2019), the grade prediction was made to identify the relationship between the multiple of inputs in the education procedure and the student's performance. Thus, the data of students used are students' attendance and grade, number of course login, and school leaving grade. Moreover, for article Dragana & Gabriela (2019), biographical essays are collected as the data used for prediction as its aim is to predict students' English grades that they had achieved for courses with Specific Purposes. Last article, Álvaro & Bruno (2019) for this objective used data that achieved from through an Analysis of logged data of online interactions.
As for the other main objective, which is predicting the students' academic score are related to articles Do, et.al. (n.d.), and Wang, et.al. (2019). the articles collect different data as article Do, et.al. (n.d.) focussed to predict score obtained from incomplete and optional courses in the third-fourth year that need data of students' grades. While article Wang, et.al. (2019), focussed on predicting score of the optional course which need the course-score records of the students to carry out the prediction.
The last objective for this subtopic is predicting the performance based on Meng, Jian, Guoxi & Kai. (2019). It is quite different from the first objective as it might not related to academic at all. To achieve this objective mentioned, data that had been used are those values that were obtained from product data management (PDM) system, which are the configuration parameters and performance parameters of each product.

Conclusions
One of the important things that need to be considered in solving the problem of Student Academic Achievement Prediction is Attributes.
The attributes used include demographics, academic grades, school-related and social attributes. The academic score attributes are the results of semester 1, semester 2 and semester 3, while the school-related attributes are high school English test results, high school mathematics results, high school physics exam results, high school biology exam results, agricultural test results, high school exams, test results Middle School Economics, High School Chemistry Exam Results. Demographic and social attributes are Gender, Mother's & Father's Education, Sibling Education, Financial problems, Advisory impact, parental reactions to low / high scores, Time spent on sports & games even on the type of play you are interested in.