Fifth, read the paper

Fifth, read the paper

• Paper name: Chen Zijian, Zhu Xiaoliang education based on academic data mining, predictive modeling online learner model.
• Study
  mining factors affecting the academic achievement of learners from online education data and build predictive classification model .
• Research motivation
  forecasting and evaluation of academic performance is a topic of common concern educational researchers around the world, how online learners on academic achievement predict facilitate the provision of early warning and other interventions.
• Review of the literature
  in this study is the use of prediction models Academic Student Performance category known training data or classification function to obtain a classification model (classifier), and evaluate the performance of the model.
• Study Design
  • Prediction algorithm
    study was forecast at the end of the study academic performance categories for the prediction of discrete values , so the use of classified strategy .
    Commonly used classification algorithm
    BN, DT, ANN, SVM (training single classifier)
  • Integrated learning
    collective decision-making is better than individual decisions (bagging, promotion, random forest)
  • Experimental design
    above four classification algorithms trained classifier in a single training set, respectively, then the four algorithms trained classifier group, and then integrated using three classifiers constituting the ensemble learning.
    Comparative performance of a single integrated classifier and classifiers.
  • The results analysis
    BN, DT, ANN, by constructing ensemble classifiers can be varying degrees of improvement classification performance, SVM is not obvious.
• Use the data set
  e-Learning University of Jordan academic data set used in the study (xAPI-Edu-Date)
• Analysis conclusion
  
• Learning experience
  a good one document, when read in conjunction with data mining chapter VIII, section 8.6 "technology to improve classification accuracy", the literature clearly shows the data set sources, but appeared to open the URL 404, in addition to the literature authors have Learn more about.