The Application of Data Mining to Build Classification Model for Predicting Graduate Employment

TL;DR

This paper develops and compares various data mining classification models, including Bayesian and Tree methods, to predict graduate employment status using real data from Maejo University.

Contribution

It introduces a graduate employment prediction model using multiple data mining algorithms and compares their effectiveness on real-world data.

Findings

Bayesian methods outperform Tree methods in accuracy.

The model accurately predicts employment status with high precision.

Different algorithms show varying performance on the dataset.

Abstract

Data mining has been applied in various areas because of its ability to rapidly analyze vast amounts of data. This study is to build the Graduates Employment Model using classification task in data mining, and to compare several of data-mining approaches such as Bayesian method and the Tree method. The Bayesian method includes 5 algorithms, including AODE, BayesNet, HNB, NaviveBayes, WAODE. The Tree method includes 5 algorithms, including BFTree, NBTree, REPTree, ID3, C4.5. The experiment uses a classification task in WEKA, and we compare the results of each algorithm, where several classification models were generated. To validate the generated model, the experiments were conducted using real data collected from graduate profile at the Maejo University in Thailand. The model is intended to be used for predicting whether a graduate was employed, unemployed, or in an undetermined situation.