Feature Selection in Data Mining using Permutation Combination

Fifie Francis, Saleema J S

Abstract


The identification of useful patterns from a medical dataset benefits early diagnosis. Classification techniques in data mining are famous for medical data prediction. Decision Tree (J48) and Navie Bayes are the most popular and frequently used algorithms in the field of prediction analysis. The efficiency of such prediction algorithms can be improved with better pre-processing approaches. This paper aims at finding optimum features using permutation combination of input data attributes for improving the classifier accuracy. The Pima Indian Diabetes Dataset from UCI repository is used for experimentation. The performance of J48 and Navie Bayes has been tested for different combination of features.


Keywords: Diabetes; J48; Navie Bayes; Permutation

Full Text:

PDF


DOI: https://doi.org/10.26483/ijarcs.v8i3.2952

Refbacks

  • There are currently no refbacks.




Copyright (c) 2017 International Journal of Advanced Research in Computer Science