大规模数据挖掘：02-Association Rules资料.pdfVIP

CS246: Mining Massive Datasets Jure Leskovec, Stanford University Supermarket shelf management 鈥 Market-basket model: 飩 Goal: Identify items that are bought together by sufficiently many customers 飩pproach: Process the sales data collected with barcode scanners to find dependencies among items 飩 classic rule: 飩 If someone buys diaper and milk, then he/she is likely to buy beer 飩 Don鈥檛 be surprised if you find six-packs next to diapers! 1/8/2014 Jure Leskovec, Stanford CS246: Mining Massive Datasets, 2 TID Items 飩 A large set of items 1 Bread, Coke, Milk 2 Beer, Bread 飩.g., things sold in a 3 Beer, Coke, Diaper, Milk supermarket 4 Beer, Bread, Diaper, Milk 飩 A large set of baskets 5 Coke, Diaper, Milk 飩 Each basket is a small subset of items 飩.g., the things one customer buys on one day 飩 A general many-to-many mapping (association) between two kinds of things 飩ut we ask about connections among 鈥渋tems鈥? not 鈥渂askets鈥 1/8/2014 Jure Leskovec, Stanford CS246: Mining Massive Datasets, 3 飩 Given a set of baskets Input: TID Items 飩 Want to discover 1 Bread, Coke, Milk 2 Beer, Bread association rules 3 Beer, Coke, Diaper, Milk 4 Beer, Bread, Diaper, Milk 飩eople who bought 5 Coke, Diaper, Milk {x,y,z} tend to buy {v,w} 飩mazon! Output: