XGBoost-PCA-BPNN prediction model and its application on predicting the effectiveness of non-surgical periodontal treatment
by Jinxiang Chen; Dong Shi; Jinqi Fan; Huanxin Meng; Jian Jiao; Ruifang Lu
International Journal of Modelling, Identification and Control (IJMIC), Vol. 34, No. 3, 2020

Abstract: An XGBoost-PCA-BPNN classification and predication model for the big data set with uncertain and coupling multidimensional characteristics is presented to solve the problem that the existing machine learning algorithms cannot get the high prediction performances. The clinical data set of the NSPT for Chinese population with CP is a typical big data set with uncertain and coupling multidimensional characteristics. The XGBoost-PCA-BPNN model is constructed to predict the effectiveness of NSPT for Chinese population with CP in this paper. The model is verified by applying it to predict the effectiveness of NSTP based on 45,000 clinical sites data set with eight characteristics. Prediction results show that the R2-score obtained by applying the Xgboost-PCA-BPNN model is 0.943, which is higher than those obtained by using Logistic regression, Xgboost, and Xgboost-Logistic regression, respectively. In addition, it is found that the effectiveness of NSTP is mainly influenced by PD, BI, sites and age.

Online publication date: Fri, 04-Dec-2020

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Modelling, Identification and Control (IJMIC):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com