迴歸分析通常被用來描述兩個或兩個以上變數間的關係，或藉由一群自變數來預測某一應變數的相關資訊。然而，通常我們只知道自變數會對應變數造成影響，至於兩者間真正的函數型態為何，卻不得而知。因此，本文試圖介紹不同型式的殘差圖，諸如：簡單殘差圖(simple residual plot)、加變數解釋圖(added-variable plot)、部份殘差圖（partial residual plot或component-plus-residual plot）、增加部份殘差圖(augmented partial residual plot)，藉由圖形所提供的資訊，希望能更有效率地找出適當的函數關係，將資料作轉換，使線性迴歸模式適用於轉換後的資料。 / The primary goal in a regression analysis is to understand how a response variable depends on one or more predictors, and to predict the value of response variable according to the predictors. However, most of the time, we only know that the predictors will have effect on the response variable, but not the true function of them. Therefore, some different forms of residual plot are considered in the study, including simple residual plot, added-variable plot, partial residual plot (or component-plus-residual plot), and augmented partial residual plot. In view of these residual plots, we can visualize easily the dependence of a response on predictors. Hence, after transforming the data using an appreciate function suggested by the plots, the data can be better fitted.