本研究使用委員會機器AdaboostM1與Bagging兩種技術方法,針對淋巴病變,提高人工智慧資料探勘之準確度。所謂淋巴癌,又稱惡性淋巴瘤(Malignant Lymphoma)。到目前為止,發生淋巴癌的原因仍不甚清楚,可能造成淋巴病變的原因有很多種。因此,本研究利用人工智慧資料探勘(Data Mining)能夠從大量資料中萃取出潛在影響因素之功能,針對罹患淋巴患者的資料進行實驗,並且建立一個淋巴分類技術模型。本研究所使用的方法包括ID3、C4.5兩種決策樹、支向機(SVM)與倒傳遞類神經網路BPN,並嘗試運用前置處理工程Resample技術並使用AdaboostM1與Bagging兩種委員會機器來提高準確度,結果並與過去相關文獻結果做比較。研究結果顯示,AdaboostM1 與Bagging能夠提高預測準確度,而使用AdaboostM1提高準確度,所獲的結果為24類結果中最佳(94.5946%)其效果比使用Bagging 來的好。
In this research, the use of two committee machine AdaboostM1 and Bagging’s techniques for lymphatic diseases, to improve the predict accuracy of artificial intelligence data mining. The so-called lymphoma, also called Malignant Lymphoma. So far, the reasons for the occurrence of lymphoma is still unclear, there are many reasons may cause lymphatic disease. Therefore, the study use data mining of artificial intelligence can extracted features of the potential impact of factors from the large amounts of data, experimenting with the data from the patients who suffering from lymphatic and then established a classification model of the lymphatic. The methods used in this research, include two decision tree ID3 and C4.5, Support Vector Machine(SVM) and back-propagation neural network(BPNN), moreover this research try to use Resample techniques in the preprocessing step, further use committee machine(AdaboostM1 and Bagging) to up the accuracy of prediction, result show will compare with the past related literatures. This research indicated AdaboostM1 and Bagging can improve the accuracy of predict, however using AdaboostM1 we got the best predict accuracy(94.5946%) than Bagging from the 24 classifications.