IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 13, No. 4, December 2024, pp. 4537~4543

ISSN: 2252-8938, DOI: 10.11591/ijai.v13.i4.pp4537-4543  4537

Multi-label feature aware XGBoost model for student

performance assessment using behavior data in online learning
environment

Shashirekha Hanumanthappa1, Chetana Prakash2

1
Department of Computer Science and Engineering, Visvesvaraya Technological University, Mysore, India
2
Department of Computer Science and Engineering, Bapuji Institute of Engineering and Technology, Davanagere, India

Article Info ABSTRACT

Article history: In light of recent outbreaks like COVID19, the use of online-based learning
streams (i.e., e-Learning systems) has increased significantly. Institutional
Received Nov 24, 2023 efforts to boost student achievement have made precise predictions of
Revised Mar 11, 2024 academic success a priority. To analyze student sessions-streams and
Accepted Mar 21, 2024 anticipate academic success, e-learning platforms are starting to combine
data mining (DM) with machine-learning (ML) techniques. Recent research
highlights the difficulties that ML-based methods have while dealing with
Keywords: unbalanced data. In tackling ensemble-learning, we combine several ML
algorithms to select the most appropriate approach for the given data.
Ensemble learning Current ensemble-based approaches for predicting student achievement,
Feature extraction nevertheless, don't do exceptionally well, particularly when it comes to
Feature selection multi-label classification, because they don't factor the relevance of features
Machine learning into their approaches. This study presents multi-label feature aware
Multi-label classification XGBoost (MLFA-XGB) method that improves upon the previously used
Student performance ensemble-learning technique. The MLFA-XGB makes use of a robust cross-
validation approach for gaining a deeper understanding of feature
relationships. The experimental results demonstrate that in comparison with
the state-of-the-art ensemble-based student achievement predictive
approach, this suggested MLFA-XGB based approach provides much higher
accuracy for prediction.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Shashirekha Hanumanthappa
Department of Computer Science and Engineering, Visvesvaraya Technological University
Mysore, India
Email: [email protected]

1. INTRODUCTION
Educational and industrial training has shifted beyond the traditional offline method towards a digital
online-mode, known as an e-learning environment [1], because of the proliferation of internet access alongside
the development of technological devices. The importance within an e-learning environment was driven
strongly throughout the COVID-19 outbreak when every school was switched to a fully online instructional
method. A trustworthy and precise approach for predicting performance among learners [2] is difficult to
provide. Academic achievement for learners can be enhanced using individualized curriculum if an efficient
evaluation approach is developed by analyzing session recordings from an online learning environment.
The biggest problems with today's e-learning systems [3] originate from the fact that they don't
allow for the sharing of material that can be customized to each student's unique preferences and learning
approach. In order to better understand each student, educators have placed an emphasis on using adaptive

Journal homepage: http://ijai.iaescore.com

4538  ISSN: 2252-8938

personalization strategies [4]. Machine-learning (ML) and data-mining (DM) techniques were just recently
put to the mission of predicting academic success for individual students. As illustrated in Figure 1, DM has
been employed to gain valuable information through the session-streams information of a particular e-
learning platform's learners, which in turn has improved decision-making and increased productivity [5]. The
applications of DM and ML in a variety of sectors, like organization, information security, and educational
opportunities [6] show great promise. Education data-mining (EDM) [7] is a relatively new discipline that
aims to improve teaching methods, learner profiles and academic outcomes [8]. Various kinds of data make
up the EDM, including records of administrative actions, records of learner session-streams operation, and
records of learner academic achievement. EDM datasets were made available in [9], [10], which gathered
information from several online resources. They used several ML models along with an ensemble-learning
technique to forecast how well students will do throughout the course of study. The results demonstrate that
the ensemble approach provides the most accurate predictions. Nevertheless, these approaches failed to
create a feature affecting prediction approach, leading to low accuracy in classifying whenever the
information considering multi-label classification and imbalanced data problem.

Figure 1. Architecture of proposed

The primary aim of this study is to develop a robust learner prediction approach which predicts
accurately student grades throughout a given course by developing multi-label feature aware (MLFA)-XGB
to effectively analyze learner session-streams e-learning information. The model is very effective in
classifying the performance of students considering three different classes such as weak, average, and good.
The proposed model achieves much higher accuracy than current ML and ensemble-based approaches.
Manuscript organization. In section 2 the detailed survey of various existing methodologies and
their limitations have been highlighted. The proposed methodology is discussed in section 3. The experiment
study using student session stream data is given in section 4. The last section significance of work is given
and future research direction for enhancing student performance prediction outcomes.

2. LITERATURE SURVEY
In the literature survey section, several studies have been discussed, each focusing on predicting
student performance using ML algorithms in various educational settings. These studies leverage different
datasets, feature selection techniques, and ML algorithms to forecast academic achievement and enhance the
quality of education. In [11], [12], the primary objective was to predict student performance at different
stages of course delivery using ML. Two separate datasets representing course delivery at 20% and 50%
completion stages were meticulously analyzed. The study begins with a feature analysis to gain insights into
the dataset's nature, which informs the selection of ML algorithms and their parameters. A systematic
approach based on the Gini index and p-value is proposed to choose a suitable ensemble learner from six
potential ML algorithms. The experimental results indicate that the proposed ensemble models achieve high
accuracy and low false positive rates for both datasets at all stages.
In [13]–[15], this research focuses on undergraduate datasets from two distinct universities and aims
to predict student achievement at two points during course delivery. It follows a similar approach to [16],

Int J Artif Intell, Vol. 13, No. 4, December 2024: 4537-4543

Int J Artif Intell ISSN: 2252-8938  4539

with the selection of ML algorithms and parameter optimization. A multi-split methodology based on the
Gini index and p-value is employed to optimize a bagging ensemble learner from six foundation ML
algorithms. The experimental findings show that the suggested bagging ensemble models deliver good
accuracy for the target group in both datasets. Figure 1 shows the proposed architecture.
The central focus of Shahzad et al. [17] was on predicting student performance during online
interactive sessions using a dataset collected from digital electronics education and design suites. The dataset
captures student interactions during online lab work, including text editing, keystrokes, time spent in
activities, and exam scores per session. The research introduces a prediction model consisting of 86 statistical
features, categorized into three broad groups: activity type, timing statistics, and peripheral activity count.
Feature selection is used to retain influential features, and five popular classifiers, including random forest
(RF) and support vector machine (SVM), are employed. The model aims to predict whether a student's
performance will be low or high. Three different scenarios for model evaluation are considered, and the
results demonstrate exceptional classification accuracy, with RF achieving the best performance at 97.4%. In
[18]–[20], this study focuses on predicting final exam grades of undergraduate students using their midterm
exam grades as source data. It employs various ML algorithms, including K-nearest neighbors (KNN), RF,
SVM, naïve Bayes (NB), and logistic regression (LR), to make predictions. The dataset comprises academic
achievement grades of 1854 students in a Turkish Language-I course. The proposed model, based on only
three parameters (midterm exam grades, department data, and faculty data), achieved a classification
accuracy of 70-75%. This study is essential for establishing a learning analysis framework in higher
education and aiding in decision-making processes, particularly for identifying students at high risk of
failure.
Pongpaichet et al. [21] introduces a ML approach to predict student performance in an online
learning environment via the Maharat platform at Taif University, following online learning training
standards in Saudi Arabia. Feature extraction is performed using hybrid optimization, and the SVM technique
is applied for predictions. The primary objective is to forecast academic achievement and assess the quality
assurance of online training programs. Descriptive-analytical methods are used to analyze sample opinions
about quality assurance. This study bridges the gap between online learning standards and student
performance prediction, contributing to enhancing the quality of online education. Several researchers [22],
[23] proposes a multi-output hybrid ensemble model that utilizes data from the superstar learning
communication platform (SLCP) to predict grades. It uses the XGB model to predict mid-term and final
grades, achieving an accuracy of 78.37%, surpassing comparison models. Additionally, the gradient-boosting
model is employed to predict homework and experiment grades, outperforming comparison models in mean
squared error. This multi-output hybrid ensemble model provides insights into how grade predictions can
improve both student learning quality and teacher teaching effectiveness [24].
In summary, these studies collectively employ ML algorithms to predict student performance and
enhance the quality of education. They leverage various datasets, feature selection techniques, and ensemble
learning methods to achieve high accuracy in predicting academic achievements, ultimately contributing to
the improvement of educational processes and outcomes. Each study offers unique insights and
methodologies, catering to different educational settings and objectives. However, considering multi-label
classification the current method exhibits poor accuracies. The proposed work is aimed at designing an
effective method for improving accuracies in performing multi-label classification.

3. MULTI-LABEL FEATURE AWARE XGBOOST MODEL FOR STUDENT PERFORMANCE

ASSESSMENT
Here, we introduce an enhanced ML approach called MLFA-XGB. Which has been developed
specifically for the purpose of EDM in the context of learner session-streams as described in Figure 2. The
MLFA-XGB algorithm represents an advancement over the conventional MLFA-XGB approach by
incorporating a more efficient selecting features process.
The XGB approach represents an enhanced iteration of the previous gradient-boosting approach
[25]. It involves the aggregation of less effective classifiers to form a robust classifier, resulting in improved
classification results. Let us consider a dataset denoted as 𝐸, that represents an ongoing stream of learning
session information. This dataset consists of 𝑜 examples, where each sample is represented by a pair (𝑦𝑗 , 𝑧𝑗 ).
Here, 𝑦𝑗 represents a vector of 𝑛 features, and 𝑧𝑗 represents a label associated with the example. The variable
𝑧̂𝑗 is utilized to denote the expected result generated by the approach in the following manner.

𝑧̂𝑗 = ∑𝐿𝑙−1 𝑔𝑙 (𝑦𝑗 ), 𝑔𝑙 ∈ 𝐺 (1)

Multi-label feature aware xgboost model for student performance … (Shashirekha Hanumanthappa)
4540  ISSN: 2252-8938

Figure 2. Architecture of proposed

The term 𝑔𝑙 refers to an independent regression-tree, while 𝑔𝑙 (𝑦𝑗 ) denotes the corresponding
prediction results generated by the 𝑙 𝑡ℎ tree for the 𝑗𝑡ℎ sample as shown in (2). The construction of any
ensemble-tree is achieved by means of a summation method. The anticipated results for the 𝑗𝑡ℎ sample during
(𝑢)
the 𝑢𝑡ℎ iteration, denoted as 𝑧̂𝑗 , necessitates the inclusion of 𝑔𝑢 in order to minimize the specified function.
The evaluation of 𝛽 is given as given in (4).

𝐺 = {𝑔(𝑦) = 𝑥𝑡(𝑦) } (2)

(𝑢−1)
𝑂(𝑢) = ∑𝑜𝑗=1 𝑚 (𝑧𝑗 , 𝑧̂𝑗 + 𝑔𝑢 (𝑦𝑗 )) + 𝛽(𝑔𝑙 ) (3)

1
𝛽(𝑔𝑙 ) = 𝛿𝑈 + 2 𝜇‖𝑥‖2 (4)

The regularization-variable is denoted by 𝛿 and 𝜇 , while the leaf's-size is represented by 𝑈 .

Additionally, the ranking for various leaves is denoted by 𝑥. The (3) can be reduced by employing the
technique of removing the stable variable using the second-order Taylor's expanding, which can be expressed
in the following manner.

1 2
𝑂(𝑢) = ∑𝑜𝑗=1 [ℎ𝑗 𝑔𝑗 (𝑦𝑗 ) + 2 𝑖𝑗 𝑔𝑢 (𝑦𝑗 ) ] + 𝛽(𝑔𝑙 ) (5)

The variable ℎ𝑗 is used to denote the initial order-gradient with regard to 𝑚, and the variable 𝑖𝑗 is used
to denote the next order-gradient with regard to 𝑚. In attaining more optimal performance with less fluctuation
the work introduces a cross-entropy loss function aware gradient boosting tree as defined in (6). The parameter
𝑧̂𝑗 is calculated in (7), and for activation sigmoid operation is computed as given in (8). Then, the work
introduces a K-fold cross validation for selecting and ranking feature with less training error using (9).

𝑀 = − ∑𝑜𝑗=1 [𝑧𝑗 𝑙𝑜𝑔(𝑧̂𝑗 ) + (1 − 𝑧𝑗 )𝑙𝑜𝑔(1 − 𝑧̂𝑗 )] (6)

Int J Artif Intell, Vol. 13, No. 4, December 2024: 4537-4543

Int J Artif Intell ISSN: 2252-8938  4541
1
𝑧̂𝑗 = [1+𝑒𝑥𝑝(−𝑎 )], (7)
𝑗

𝜕𝑧̂ 𝑗
= 𝑧̂𝑗 (1 − 𝑧̂𝑗 ) (8)
𝜕𝑎𝑗

1
𝐶𝑉(𝜎) = 𝑆𝑀 ∑𝑆𝑠=1 ∑𝐾 ̂𝜎−𝑘(𝑗) (𝑦𝑗 , 𝜎))
𝑘=1 ∑𝑗∈𝐺−𝑘 𝑃 (𝑏𝑗 , 𝑔 (9)

In (9), to select ideal 𝜎̂ for optimizing the student prediction model is attained as follows. Where 𝑀
defines size of training dataset considered, 𝑃(∙) defines loss function, and 𝑔̂𝜎−𝑘(𝑗) (∙) defines a function to
compute coefficients. The proposed MLFA-XGB based student performance prediction model achieves better
classification accuracy in comparison with existing ensemble-based classifier as shown in result section.

𝜎̂ = 𝐶𝑉𝑠 (𝜎) (10)

4. RESULT AND ANALYSIS

This section delves into the examination of student performance prediction by employing the
presented MLFA-XGB approach alongside additional established ML-based approaches for learner
prediction [10], [15]. The evaluation of performance in this study utilizes the e-learning dataset obtained
using [10]. The dataset preference is predicated upon the findings presented in a comparative study [9], [10],
[15]. The ML approach utilized in this study for the purpose of predicting student performance has been
developed with the Python 3 framework. The accuracy, specificity, sensitivity, and F1-score are metrics used
for validating models. The proposed work namely MLFA-XGB is compared with existing methodologies
namely multi-split optimization bagging ensemble (MSOBE) [10], RF-ensemble [15], and XGB [9], [10].
The specificity performance is given in Figure 3. The results show that the MSOBE achieves much
less specificity, the XGB model achieves better performance than RF-ensemble and MSOBE. On the other
side, the proposed MLFA-XGB achieves much better specificity performance than other existing student
performance classification methods. The sensitivity performance is given in Figure 4. The results show that
the MSOBE achieves much less sensitivity, the XGB model achieves better performance than RF-ensemble
and MSOBE. On the other side, the proposed MLFA-XGB achieves much better sensitivity performance than
other existing student performance classification methods.

Specificity Sensitivity
1.2 RF-Ensemble MSOBE XGB MLFA-XGB 1.2 RF-Ensemble MSOBE XGB MLFA-XGB

1 1
Specificity

0.8 0.8
Sensitivity

0.6 0.6

0.4 0.4

0.2 0.2

0 0
RF-Ensemble MSOBE XGB MLFA-XGB RF-Ensemble MSOBE XGB MLFA-XGB
Prediction Models Prediction Models

Figure 3. Specificity performance Figure 4. Sensitivity performance

The accuracy performance is given in Figure 5. The results show that the MSOBE achieves much
less accuracy, the XGB model achieves better performance than RF-ensemble and MSOBE. On the other
side, the proposed MLFA-XGB achieves much better accuracy performance than other existing student
performance classification methods.
The F1-score performance is given in Figure 6. The results show that the MSOBE achieves much
less F1-score, the XGB model achieves better performance than RF-ensemble and MSOBE. On the other
side, the proposed MLFA-XGB achieves much better accuracy performance than other existing student
performance classification methods.

Multi-label feature aware xgboost model for student performance … (Shashirekha Hanumanthappa)
4542  ISSN: 2252-8938

Accuracy
1.2 RF-Ensemble MSOBE XGB MLFA-XGB

0.8

Accuracy
0.6

0.4

0.2

0
RF-Ensemble MSOBE XGB MLFA-XGB
Predicitve Model

Figure 5. Accuracy performance

F1-Score
1.2
RF-Ensemble MSOBE XGB MLFA-XGB
1

0.8
F1-Score

0.6

0.4

0.2

0
RF-Ensemble MSOBE XGB MLFA-XGB
Predicitve Model

Figure 6. F-measure performance

5. CONCLUSION
The present study introduces a novel ensemble ML approach which demonstrates enhanced
efficiency through the modification of the XGBoost algorithm. Notably, this method exhibits robust
performance even in scenarios where the training information suffers from imbalanced class distribution. In
this study, we offer a novel and efficient cross-validation method that aims to determine the specific features
that have significant effects on the correctness of a prediction approach. The utilization of the CV method
involves the implementation of a proficient feature ranking method, which aims to enhance the accuracy of
predictions by minimizing the prediction-error. The research study was carried out utilizing a dataset
consisting of conventional student-session streaming information. The MLFA-XGB method exhibits notable
enhancements in terms of precision, accuracy, specificity, sensitivity, and F-measure performance when
compared to existing student performance predictive approaches using RF-ensemble, MSOBE, and XGB-
based approaches. The future work would be focused in enhancing the model further and also further validate
the model under more diverse dataset.

REFERENCES
[1] A. E. Tatar and D. Düştegör, “Prediction of academic performance at undergraduate graduation: Course grades or grade point
average?,” Applied Sciences, vol. 10, no. 14, 2020, doi: 10.3390/app10144967.
[2] A. Moubayed, M. Injadat, A. Shami, and H. Lutfiyya, “Student engagement level in an e-learning environment: clustering using
k-means,” American Journal of Distance Education, vol. 34, no. 2, pp. 137–156, 2020, doi: 10.1080/08923647.2020.1696140.
[3] S. A. Priyambada, T. Usagawa, and M. ER, “Two-layer ensemble prediction of students’ performance using learning behavior and
domain knowledge,” Computers and Education: Artificial Intelligence, vol. 5, 2023, doi: 10.1016/j.caeai.2023.100149.
[4] A. Kumar, A. Kaur, P. Singh, M. Driss, and W. Boulila, “Efficient multiclass classification using feature selection in high-
dimensional datasets,” Electronics, vol. 12, no. 10, 2023, doi: 10.3390/electronics12102290.
[5] A. Al-Zawqari, D. Peumans, and G. Vandersteen, “A flexible feature selection approach for predicting students’ academic
performance in online courses,” Computers and Education: Artificial Intelligence, vol. 3, 2022, doi: 10.1016/j.caeai.2022.100103.
[6] K. Jawad, M. A. Shah, and M. Tahir, “Students’ academic performance and engagement prediction in a virtual learning
environment using random forest with data balancing,” Sustainability, vol. 14, no. 22, 2022, doi: 10.3390/su142214795.

Int J Artif Intell, Vol. 13, No. 4, December 2024: 4537-4543

Int J Artif Intell ISSN: 2252-8938  4543

[7] P. Pujar, A. Kumar, and V. Kumar, “Plant leaf detection through machine learning based image classification approach,” IAES
International Journal of Artificial Intelligence, vol. 13, no. 1, pp. 1139–1148, 2024, doi: 10.11591/ijai.v13.i1.pp1139-1148.
[8] S. H. Sreedhara, V. Kumar, and S. Salma, “Efficient big data clustering using adhoc fuzzy c means and auto-encoder CNN,”
Lecture Notes in Networks and Systems, vol. 563, pp. 353–368, 2023, doi: 10.1007/978-981-19-7402-1_25.
[9] M. N. Injadat, A. Moubayed, A. B. Nassif, and A. Shami, “Systematic ensemble model selection approach for educational data
mining,” Knowledge-Based Systems, vol. 200, 2020, doi: 10.1016/j.knosys.2020.105992.
[10] M. N. Injadat, A. Moubayed, A. B. Nassif, and A. Shami, “Multi-split optimized bagging ensemble model selection for multi-
class educational data mining,” Applied Intelligence, vol. 50, no. 12, pp. 4506–4528, 2020, doi: 10.1007/s10489-020-01776-3.
[11] M. L. Nistal, “An experience of continuous assessment in telecommunication technologies engineering: New costs for the
teacher,” Revista Iberoamericana de Tecnologias del Aprendizaje, vol. 8, no. 2, pp. 90–95, 2013, doi:
10.1109/RITA.2013.2258225.
[12] M. Yağcı, “Educational data mining: prediction of students’ academic performance using machine learning algorithms,” Smart
Learning Environments, vol. 9, no. 1, 2022, doi: 10.1186/s40561-022-00192-z.
[13] M. N. Alsubaie, “Predicting student performance using machine learning to enhance the quality assurance of online training via
Maharat platform,” Alexandria Engineering Journal, vol. 69, pp. 323–339, 2023, doi: 10.1016/j.aej.2023.02.004.
[14] H. Xue and Y. Niu, “Multi-output-based hybrid integrated models for student performance prediction,” Applied Sciences, vol. 13,
no. 9, 2023, doi: 10.3390/app13095384.
[15] G. B. Brahim, “Predicting student performance from online engagement activities using novel statistical features,” Arabian
Journal for Science and Engineering, vol. 47, no. 8, pp. 10225–10243, 2022, doi: 10.1007/s13369-021-06548-w.
[16] E. Alhazmi and A. Sheneamer, “Early predicting of student’s performance in higher education,” IEEE Access, vol. 11, pp. 27579–
27589, 2023, doi: 10.1109/ACCESS.2023.3250702.
[17] R. Shahzad et al., “Multi-agent system for student’s cognitive assessment in e-learning environment,” IEEE Access, vol. 12, pp.
15458–15467, 2024, doi: 10.1109/ACCESS.2024.3356613.
[18] Z. Xu, H. Yuan, and Q. Liu, “Student performance prediction based on blended learning,” IEEE Transactions on Education, vol.
64, no. 1, pp. 66–73, 2021, doi: 10.1109/TE.2020.3008751.
[19] P. Jiang and X. Wang, “Preference cognitive diagnosis for student performance prediction,” IEEE Access, vol. 8, pp. 219775–
219787, 2020, doi: 10.1109/ACCESS.2020.3042775.
[20] N. A. Butt, Z. Mahmood, K. Shakeel, S. Alfarhood, M. Safran, and I. Ashraf, “Performance prediction of students in higher
education using multi-model ensemble approach,” IEEE Access, vol. 11, pp. 136091–136108, 2023, doi:
10.1109/ACCESS.2023.3336987.
[21] S. Pongpaichet, K. Nirunwiroj, and S. Tuarob, “Automatic assessment and identification of leadership in college students,” IEEE
Access, vol. 10, pp. 79041–79060, 2022, doi: 10.1109/ACCESS.2022.3193935.
[22] J. Figueroa-Canas and T. Sancho-Vinuesa, “Early prediction of dropout and final exam performance in an online statistics
course,” Revista Iberoamericana de Tecnologias del Aprendizaje, vol. 15, no. 2, pp. 86–94, 2020, doi:
10.1109/RITA.2020.2987727.
[23] J. L. P.-Lujan, C. T. Calafate, J. L. P.-Yague, and J. C. Cano, “Assessing the impact of continuous evaluation strategies: tradeoff
between student performance and instructor effort,” IEEE Transactions on Education, vol. 59, no. 1, pp. 17–23, 2016, doi:
10.1109/TE.2015.2418740.
[24] A. Smith, S. L.-Munk, A. Shelton, B. Mott, E. Wiebe, and J. Lester, “A multimodal assessment framework for integrating student
writing and drawing in elementary science learning,” IEEE Transactions on Learning Technologies, vol. 12, no. 1, pp. 3–15,
2019, doi: 10.1109/TLT.2018.2799871.
[25] T. Chen and C. Guestrin, “XGBoost: a scalable tree boosting system,” Proceedings of the ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining, vol. 13-17-Augu, pp. 785–794, 2016, doi: 10.1145/2939672.2939785.

BIOGRAPHIES OF AUTHORS

Mrs. Shashirekha Hanumanthappa currently working as Asst. Professor in the

Department of Computer Science and Engineering, Visvesvaraya Technological University
Centre for Post Graduation Studies, Mysuru. She has completed M.Tech. in Computer Science
and Engineering from UBDT College of Engineering (Kuvempu University), Davanagere,
Karnataka, India in the year 2008. Her field of interest is big data, artificial intelligence, and
machine learning. She can be contacted at email: [email protected].

Dr. Chetana Prakash holds Doctor of Philosophy (Ph.D.) in Computer Science

and Engineering and she is currently working as Professor in the Department of Computer
Science and Engineering, Bapuji Institute of Engineering and Technology, Davangere. She has
teaching experience of more than 30 Years. Her field of interest is speech signal processing,
data mining, image processing, fuzzy techniques, IoT, and data analytics. She can be contacted
at email: [email protected].

Multi-label feature aware xgboost model for student performance … (Shashirekha Hanumanthappa)