碩士 / 國立屏東商業技術學院 / 資訊管理系 / 94 / Though the GNP of Taiwan has already been up to 16,000 US dollars, it has accorded with the standard that World Bank regards as a rich country, but child abuse incidents increase year by year. It is worth while studying how to predict the occurrence of the child abuse in advance to prevent consequent tragedy. Despite of a lot of researches have discussed the question of child abuse, but few studies probe into the marital violence and the child abuse simultaneously. This study builds a predication model of child abuse caused by marital violence based on sociodemograpghy and marital violence factor, which is analyzed by Logistic Regression, Back-propagation Neural Networks (BPN) and Support Vector Machine (SVM) model. The prediction models were verified and assessed via 5-fold cross-validation and paired t-test. The misclassification errors are analyzed in terms of Type I and Type II. In addition, a ROC curve is provided to evaluate the prediction ability. Finally, experimental result of the research showed that the prediction ability of SVM model is the best than the model of Logistic Regression and BPN. We used SVM model to build a child abuse prediction’s expert system, which is useful to predict and prevent child abuse.
| Identifer | oai:union.ndltd.org:TW/094NPC05396008 |
| Date | January 2006 |
| Creators | Huai-shian Huang, 黃懷賢 |
| Contributors | Ming-chi Lee, 李明錡 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 52 |
Page generated in 0.01 seconds