| Summary: | 碩士 === 東吳大學 === 企業管理學系 === 104 === There are many product introduction apps currently on the market to attract users to download. However, it’s good for users if we can offer relative product recommendation because it would solve the information overload. In research motive, collect users two car compare records from car information app. By association rules, find out the association of cars which helps app developer to establish recommendation system. In research purpose, when user chooses car, system can show another car product data as reference. The study is from 2015/1/30 to 2015/4/2. It included 40 car brands and 870 car types. We separate user compare records to two units. (1) Study 1: two car compare records are 30,867. (2) Study 2: user compare lists are 5,327. In research, we offer width (quantity of cars which have associate products) and average depth (each car with quantity of associate) to evaluate the results of different threshold. We find (1) Support adjustment is influence on width. (2) Confidence adjustment under threshold (study1 is 10%; study2 is 20%) don’t impact width, but main point is average depth. (3) Under proper threshold, both study 1 and study 2 width are alike, but study 2 average depth is much deeper. (4) Study 1 is much accurate for compare two car's information. Study 2 offers more quantity of association rules. Finally, according to the results, the managerial implication, limitation of research and future research directions are discussed.
|
|---|
