本論文的目的是想從人們的穿著上,分析出服裝之間的關連性,找到一個合適的搭配模型來概括全部人的穿著,將這個模型提供給下一位具有相同行為的人,給予建議跟提供不同的模式,可以給他們更多的選擇。 這次的研究方向,是希望透過資料探勘(Data Mining)的方法,找出資料的聚合關係與組合關係。使用概念絡(Concept Lattice)將資料做一個基礎的分析跟分析最基本的關係。 本實驗是透過Apriori演算法,將服裝搭配風格的學院風、甜美風、混搭風、氣質風、個性風等五種風格,歸納整理出其個別的獨特規則。再利用這些規則去分類服飾為五種風格裡的其中一種風格。測試結果發現,我們所輸入的資料,正確率已達70.27%,也就是有七成的資料可由我們所歸納的規則來判斷出正確的風格。 風格可說明其隱含著的某些意義,像是適用的人與實用的時間和地點都可由風格來代表,所以只要知道所想要的風格,就能從中找到適合的服飾而組成好的搭配。所以我們以風格為主,再從風格中的每個型式中找出一項單品,搭配成套,譬如:S=甜美可愛風搭配=(白色上衣,墨綠色長褲,咖啡色紳士踝靴,黑色手拿包,紫色珠珠項鍊)就成了一套甜美可愛風格的搭配。
In this research, the relationships among costumes from people's dressing are analyzed for one suitable model to summarize all people's dressing. Then this model is provided to the next person who have same behavior, and suggestions for different model can give them more choices. In this research, The Apriori algorithm in Data Mining is used to find out the relationships among materials. Concept Lattice is used to do a basic analysis among fundamentally. In the experiment, we use Apriori to find out unique rules from costumes in five kinds of styles, and use these rules to group costumes in one of five styles. The test results show that the classification rate already been 70.27%, in other words, there are 70% materials can get the correct styles. The style can explain some implicit meanings, style can represent the using time, using place and suitable person of costume. So, if you know what kind of style you want, you can find out the suitable costume. So, we major on style, and choice one of costume in each costume type, and make a complete set. For example, white jacket, blackish green trousers, coffee gentleman's ankle boots, black bag and purple pearl necklace, they can make a sweet and cute style with complete set.