Title: Recommender System for Selection of the Right Study Program for Higher Education Students
Summary
A third way of building recommender systems in RapidMiner is shown in Chapter 10, where classification algorithms are used to recommend the best-fitting study program for higher-education students based on their predicted success for different study programs at a particular department of a particular university. The idea is an early analysis of students’ success on each study program and the recommendation of a study program where a student will likely succeed. At this university department, the first year of study is common for all students. In the second year, the students select their preferred study program among several available programs. The attributes captured for each graduate student describe their success in the first-year exams, their number of points in the entrance examination, their sex, and their region of origin. The target variable is the average grade of the student at graduation, which is discretized into several categories. The prediction accuracy of several classification learning algorithms, including Naive Bayes, Decision Trees, Linear Model Tree (LMT), and CART (Classifications and Regression Trees), is compared for the prediction of the student’s success as measured by the discretized average grade. For each student, the expected success classes for each study program is predicted and the study program with the highest predicted success class is recommended to the student. An optimization loop is used to determine the best learning algorithm and automated feature selection is used to find the best set of attributes for the most accurate prediction. The RapidMiner processes seamlessly integrate and compare learning techniques implemented in RapidMiner with learning techniques implemented in the open source data mining library Weka, thanks to the Weka extension for RapidMiner that seamlessly integrates all Weka learners into RapidMiner.
Table of Contents
10.1 Introduction
10.2 Literature Review
10.3 Automatic Classification of Students using RapidMiner
10.3.1 Data
10.3.2 Processes
10.3.2.1 Simple Evaluation Process
10.3.2.2 Complex Process (with Feature Selection)
10.4 Results
10.5 Conclusion
10.5 Bibliography
Dataset & Processes: Click here to download