地理空間資料的分析如果是以眼睛的視覺化分析為主時，資料的總計方式將對成果的解釋說明有顯著的差異。其原因可能是由著名的Modifiable Areal Unit Problem (MAUP: Openshaw, 1996)所導致，這個問題在土地使用現況的分析時尤其嚴重。因為土地使用活動在空間的分佈是有階層性的，對土地使用型態的描述就必須說明其空間尺度，不同空間尺度下的土地使用型態就會有所不同。本研究以1994年台灣地區國土利用調查中高雄地區的成果為資料基礎，使用GAM的點空間模型分析方法(GAM point spatial pattern)來解決如何導出連續性空間尺度之土地使用型態的問題。解決方案是建立以個體資料為基礎的土地使用空間圖，然後利用GAM的搜尋演算法來探索整個都市地理空間不同的尺度的商業土地使用型態，然後再加以視察化的成果展現不同尺度探索的結果。

In spatial interpretation, the way data are aggregated will result significant visualization difference if analysis of geographic phenomena is based on eyeballs. That might be due to the well-known Modifiable Areal Unit Problem. It is particularly true in interpreting land use maps Land-use patterns are multi-scales as the distribution of land use activities in space is hierarchical. To address the problem clearly, Kaohsiung commercial land use maps of various spatial levels are presented using 1994 Taiwan Area National Land Use Survey data. And GAM point spatial pattern analysis is applied to solve the problem. In which, commercial land use spatial patterns will be explored based on disaggregated unit of land parcel to analyse spatial differentiate in study area. The discussion covers MAUP and the GAM approach to retrieve information from point data models. Nonetheless, the emphasis is on the empirical study of geographical spatial data mining by exploring commercial patterns in the study area of Kaohsiung City.