資料探勘中的分類技術經常被使用於處理各種分類問題，如何從眾多的分類技術中選擇合適的方法進行分析研究即成為一個重要的課題。以往大多數的學者對於分類器性能的評估，通常著重於比較分類器的預測正確率或模型訓練的速度等等。然而，在實務上，不同的分類問題皆有其獨特的資料結構，因此可能影響著分類器的表現。本研究使用了十五個資料複雜度指標(data complexity index)以量化分類問題的資料特徵，並對於此十五個資料複雜度指標進行因素分析，探索指標之間的重複性、相關性，將選出的因素當成此十五種資料複雜度指標的綜合指標。 本文考慮了分類正確率的比例來評估一個分類器可否有效區分不同類別資料的能力。本研究的目的即是探索資料複雜度指標之間的相關性，並觀察資料特性的複雜程度對於各種分類技術的影響，研究結果也顯示，資料複雜度確實對於分類器的表現有所影響。本研究希望可以有效地提供資訊，使研究者面對一分類資料時，從資料複雜度指標值以及因素值可以預先推估可能的分類結果，也使研究者經由資料複雜度指標值或因素值，進而選擇對於欲分類的資料最適當之分類器。 Classification techniques in data mining are often used to deal with a variety of classification problems. Choosing suitable methods for analysis from many classification techniques becomes an important issue. For the performance evaluations of the classifiers, researchers used to compare them on several datasets in terms of classification accuracy or training time, and so on. In practice, however, different classification problems has their unique data complexities which might affect the accuracies of the classifiers. Therefore, we adopt fifteen data complexity indices to quantify the data characteristics and use correct classification rate to observe the influence of these indices on seven commonly used classification techniques. We also use factor analysis to explore the correlation among these indices. The results show that different data characteristics indeed have impacts on classification performance. According to our studies, for classification problems, researchers can calculate the data complexity indices or factor values suggested in this paper to estimate the classification difficulties, and also choose the most appropriate classification method on their study.
