Data Mining in Education System – An Application

Data mining helps to mine the unique & significant data from the data warehouse. It is used to express knowledge discovery and search for essential relationships among different attributes in the data warehouse. Data mining, today is used in lots of different real world applications like banking, medical, telecommunications, fraud detection, education etc. The need for Educational Data Mining (EDM) is emerged to accurately analyse the performance of each student in a pool of thousands of students taking different courses in a university environment. It also provides a way to accurately predict the future academic performance of each student based on his/her previous records. It would be extremely time consuming and inconvenient if one were to do this analysis manually and the results may not be accurate. The framework for predicting student academic performance using Data Mining techniques is given below:

EDM is carried out using computer based algorithms to identify patterns from a huge educational database. Different types of algorithms & techniques like Classification, Regression, Association Rules Mining, Genetic Algorithm, Clustering, Nearest Neighbors Method, and Decision Trees are used for information retrieval from educational databases. These techniques are further used for predicting student performance, educational dropout, provides inputs for designing new curriculum etc. A brief introduction of some of the different data mining techniques are as follows:

Classification: This technique is totally based on machine learning. It classifies each dataset into predefined classes. To classify data in database, mathematical techniques similar to neural network, decision trees, statistics and linear programming are used. Let us assume a University in which a large number of students are studying. With the help of classification techniques we can classify students based on different attributes.

Clustering: Clustering is used to make clusters of comparatively identical cases or observations. Things in a cluster are comparable with each other. They are also unrelated to things outside the cluster. We can use this technique to group different students according to their attributes like course, grade, activity, age, gender, hosteller, day-scholar, rural, urban etc. With this we can provide different types of facility to different clusters according to their specific requirements.

Prediction: Prediction is useful to predict the value of an unknown attribute with the help of some known attributes. With the help of prediction techniques, we can predict the future learning habits of the student. By applying these techniques on the academics data; we can predict the student’s performance in near future.

Association Rule mining: It is well-known and leading technique in data mining. With this technique we can find the hidden patterns between different attributes of a single dataset. This technique is also known as relational data mining technique. The primary use of it is in market-basket analysis. In educational context, it is used to find the association between different attributes of a student that affects the performance.

P. Sunil Kumar, D. Jena et al published a paper entitled “Mining the factors affecting the high school dropouts in rural areas”. In this paper authors considered seven different attributes to find out the relationship between attributes which affect students to dropout. In their analysis, they found out students who are not interested in the studies are most likely to dropout as compared to students who experience bad teaching environment and/or suffer from poverty.

By: Mukesh Thakur - Assistant Professor, CSE, Chitkara University, H.P.

References:

1. Baker, R. S. J. D., & Yacef, K. (2009). The state of educational data mining in 2009: A review and future visions. Journal of Educational Data Mining, 1, 3-17.
2. Siti Khadijah Mohamad, Zaidatun Tasi presented “Educational data mining: A review”. The 9th International Conference on Cognitive Science Procedia - Social and Behavioural Sciences 97 (2013) 320 – 324
3. Cristobal Romero and Sebastian Ventura presented “Data mining in education”. WIREs Data Mining Knowledge Discovery 2013, 3: 12–27 doi: 10.1002/widm.1075
4. Azwa Abdul Aziz, Nor Hafieza Ismailand Fadhilah Ahmad, First Semester Computer Science Students’ Academic Performances Analysis by Using Data Mining Classification Algorithms, Proceeding of the International Conference on Artificial Intelligence and Computer Science(AICS 2014), 15 - 16 September 2014, Bandung, INDONESIA. (e-ISBN978-967-11768-8-7).