Source-linked AI summary
Application of k Means Clustering algorithm for prediction of Students Academic Performance
O. J. Oyelade, O. O. Oladipupo, I. C. Obagbuwa
TL;DR
Monitoring students’ academic performance is difficult when grouping relies only on average scores, motivating a clustering-based analysis of performance progression. The paper combines k-means clustering with a deterministic model and applies it to university results from a private institution in Nigeria, producing performance groupings used as a benchmark for academic monitoring.
Problem
Traditional grouping by average scores makes it difficult to obtain a comprehensive and time-sensitive view of students’ academic performance progression.
Method
The paper combines k-means clustering with a deterministic model to analyze students’ scores and group performance levels.
Results
53.03% corresponded to “Good” performance in the reported performance index, while the analysis also identified “Fair,” “Good,” and “Very Good” performance groups.
Takeaways & Limitations
The clustering algorithm serves as a benchmark for monitoring students’ performance progression and supporting academic planners’ semester-by-semester decision making.
Abstract
from arXiv · showhide
The ability to monitor the progress of students academic performance is a critical issue to the academic community of higher learning. A system for analyzing students results based on cluster analysis and uses standard statistical algorithms to arrange their scores data according to the level of their performance is described. In this paper, we also implemented k mean clustering algorithm for analyzing students result data. The model was combined with the deterministic model to analyze the students results of a private Institution in Nigeria which is a good benchmark to monitor the progression of academic performance of students in higher Institution for the purpose of making an effective decision by the academic planners.
I. INTRODUCTION
Monitoring students’ academic performance is important for evaluating progression, but traditional score-based grouping makes it difficult to capture comprehensive and time-varying performance patterns. The paper presents k-means clustering as a simple tool for monitoring performance progression.
- I. INTRODUCTION: GPA is a commonly used indicator for evaluating students’ academic performance and progression in higher education.Academic planners use GPA to assess progression, while institutions may set minimum GPA requirements.
- I. INTRODUCTION: Grouping students by average scores alone makes it difficult to obtain a comprehensive view of performance and discover changes over time.The paper identifies this as a complicated performance-monitoring task.
- I. INTRODUCTION: Clustering methods can identify key characteristics in students’ performance data and potentially support future prediction.The paper notes prior use of k-means with Euclidean distance to measure differences between scores.
- I. INTRODUCTION: The paper presents k-means clustering as a simple and efficient tool for monitoring students’ performance progression in higher institutions.This is the paper’s stated contribution in the introduction.
- I. INTRODUCTION: Cluster-analysis methods include hierarchical and non-hierarchical techniques, and algorithm selection depends on the data and analytical purpose.The paper also identifies cluster stability and separation quality as relevant evaluation considerations.
II. METHODOLOGY
The methodology formulates k-means as an optimization problem that assigns data points to centroids while minimizing mean squared Euclidean distance. It iteratively updates centroids until a local minimum, with computational cost depending on dataset size, cluster count, and iteration count.
- II. METHODOLOGY: The method seeks k cluster centroids in R^d that minimize the average squared Euclidean distance between each data point and its nearest centroid.This objective is described as the mean squared error (MSE) of the clustering.
- II. METHODOLOGY: Euclidean distance d(x_i, m_j) measures the distance between data point x_i and centroid m_j.The distance is used within the clustering objective.
- II. METHODOLOGY: K-means approximately solves the clustering objective through simple, scalable, fast-converging, and sparse-data-adaptable computations.The paper gives these properties as reasons for k-means’ popularity.
- II. METHODOLOGY: The algorithm initializes centroids, repeatedly updates them to decrease the objective, and converges to a local minimum.The particular local minimum depends on the starting centroids, while finding the global minimum is NP-complete.
- II. METHODOLOGY: O(nkl) is the stated time complexity, where n is the number of data objects, k the number of clusters, and l the number of iterations.The iteration count varies with the initial cluster centroids, even on the same dataset.
- II. METHODOLOGY: The traditional procedure accepts the dataset and desired cluster count, then initializes the first k clusters using initial instances or random sampling.The paper presents generalized and traditional k-means pseudocode in Figures 1 and 2.
III. RESULTS
The model applies k-means clustering to one semester of Nigerian university results, representing students by course-score matrices and evaluating cluster performance with a deterministic model. Reported performance varies across cluster sizes and k settings, spanning Fair, Good, and Very Good categories.
- Data and evaluation: The dataset contains one semester of academic results from a Nigerian university, represented as an N by M matrix of students and courses.N denotes the number of students, while M denotes the courses taken by each student.
- Results: For k = 3, cluster sizes 25, 15, and 29 have overall performances of 62.22, 45.73, and 53.03, respectively.The reported 53.03% result is classified as Good performance in the performance index.
- Data and evaluation: The deterministic model evaluates each cluster by summing the averages of individual student scores within the cluster.The paper defines N as the total number of students in a cluster and n as the data dimension.
- Results: For k = 4, cluster sizes 24 and 16 have overall performances of 50.08% and 65.00%, classified as Good and Very Good, respectively.Cluster size 30 has 58.89% and cluster size 9 has 43.65%, classified as Good and Fair.
- Results: For k = 5, cluster sizes 19, 17, 9, 14, and 20 have overall performances of 49.85%, 60.97%, 43.65%, 64.93%, and 55.79%.The paper classifies these results as Good, Very Good, Fair, Very Good, and Good, respectively.
IV. DISCUSSION AND CONCLUSION
The paper proposes k-means clustering combined with a deterministic model to analyze students’ academic results. Using private-school results from Nigeria, it presents the approach as a benchmark for monitoring performance progression and supporting academic-planning decisions.
- Discussion and conclusion: The study combines k-means clustering and Euclidean-distance similarity with a deterministic model to analyze student results.The dataset contains nine courses for each of 79 students in one semester.
- Discussion and conclusion: The resulting methodology provides a simple, qualitative comparison of clustering algorithms’ predictive power and Euclidean distance as a similarity measure.The authors state that the model improves on limitations of several earlier approaches.
- Discussion and conclusion: The clustering algorithm is presented as a benchmark for monitoring students’ performance progression in higher institutions.The authors also connect it with academic planners’ semester-by-semester decision making.