Contrast Research of Two Kinds of Integrated Sorting Algorithms
Xiao Bo Li · Advanced materials research · 2012
Boosting and Bagging are two kinds of important voting sorting algorithms. Boosting algorithm can generate multiple classifiers by serialization through adjustment of sample weight; Bagging can generate multiple classifiers by parallelization. Different algorithms are composed of different loss and different integration mode, through integration of Bagging and Boosting algorithm and naïve Bayes algorithm, the Bagging NB and AdaBoost NB algorithms are constructed. Through experiment contrast of UCI data set, the result shows Bagging NB algorithm is relatively stable, it can produce the sorting result superior than that of NB algorithm, AdaBoost NB algorithm is greatly affected by the singular value in data distribution, the result with foundation of NB algorithm is relatively poor on part of data set, and that might have negative influence on the classifier algorithm.