K.j. Kim, H. Ahn Expert Systems with Applications 34(2008)1200-1209 ·必☆色·品回·回洛 Product Recommender Sysfem using Datamining 0HEA呈9平9属世智200形 装(电模登)[19形 是创⊙go 为出例以刀 ”叫种但啊:选栏口 :能2料 器口斗钻是备0驻? 日动 己实智u□g 4( ahmo//im. 8日”也 Product Recommender System using Datamining 08空用苦量空A0阳本型D0后厢量平附A 0上宝异封型2 留q♂召斗B吉q剖昝雙引睏因舁昌e譽凵? 器甚器器。等啡闻恕些垇”詈磔詈 测但器⊙O鲤⊙某O沿甘⊙异O咎 Fig. 5. Sample screens of the prototype system(a)The input screen(b) The result screen for recommendation. SOM, and GA K-means all together, and focused on finding Table 6 the algorithm which produced the best result. We used The result of the survey intraclass inertia as a performance measure. Table 2 shows Recommendation method Average Standard deviation the summarized performance of the three methods. As you Randomly-chosen model 3.7 342 can see, GA K-means is the best among three comparative Proposed model ( GA K-means CBR) 4.51 1.010 methods for this data set To determine whether the differences of intraclass inertia between the proposed model and other comparative mod- for paired samples. As explained in Section 2.3, intraclass els are statistically significant or not, we applied the t-test inertia is the mean of the distances between the clusterSOM, and GA K-means all together, and focused on finding the algorithm which produced the best result. We used intraclass inertia as a performance measure. Table 2 shows the summarized performance of the three methods. As you can see, GA K-means is the best among three comparative methods for this data set. To determine whether the differences of intraclass inertia between the proposed model and other comparative mod￾els are statistically significant or not, we applied the t-test for paired samples. As explained in Section 2.3, intraclass inertia is the mean of the distances between the cluster Fig. 5. Sample screens of the prototype system (a) The input screen. (b) The result screen for recommendation. Table 6 The result of the survey Recommendation method Average Standard deviation Randomly-chosen model 3.76 1.342 Proposed model (GA K-means + CBR) 4.51 1.010 K.-j. Kim, H. Ahn / Expert Systems with Applications 34 (2008) 1200–1209 1207