Data Mining Association Mining - BPD2233 - PDF

BPD2233: DATA MINING Association Mining 1 Learning Objectives ▪ To comprehend the concept of association mining and its applications ▪ To understand the role of support, confidence and lift 2 쇼...

BPD2233: DATA MINING Association Mining 1 Learning Objectives ▪ To comprehend the concept of association mining and its applications ▪ To understand the role of support, confidence and lift 2 쇼핑 가자 https://www.youtube.com/watch?v=oStp6NtM6JM 3 Introduction Introduction ▪ Wallmart: People who bought diaper also buy beer. Therefore, they placed beer beside the diapers rack and it increased beers’ sales ▪ WHY? Analysis could lead to “Young couple prepare for the weekend by stocking up on diapers for infants and beer for dad. Introduction Defining Association Rule Mining ▪ Defined as identification of frequent patterns, correlations, associations or casual structures among sets of a objects or item in transactional databases, relational databases and other information repositories ▪ Generally, if/then statement i.e.: “If customer buys a dozen eggs, he is 80% likely to also purchase milk” ▪ Association rule consists of 2 parts Antecedent (if): is an object or item found in the data Consequent (then): an object or item found in the combination with the antecedent ▪ Written as X---→ Y, whenever X appears Y alto tends to appear. X is antecedent and Y is consequent ▪ Formulated by 2 Indian Scientists Dr. Rakesh Agrawal and Dr R. Srikant Representation of Item Association Rule Mining ▪ Assume the number of items in the shop stock in n ▪ Six items in stock – bread, milk, diapers, beer, egg, and cola ▪ The item list represent I and its items represent by {i1,i2,……in} ▪ The number of transaction represented by N ▪ Each transaction denoted by T {t1,t2….tN) with unique identifier (TID) ▪ Task – to find association relationship given a large number of transactions, such that items that tend to occur together are identified The Metric to Evaluate the Strength of Association Rule Support ▪ Support: ▪ Calculate Support (Bread) ▪ Calculate Support (Bread, Milk) SUPPORT is important metric because if the rule has low support then it maybe the case that the rule occur by chance and it will not be logical to promote items that customers seldom by together If the rule has high support then that association become important and if implemented properly will result the increase revenue, efficiency and customer satisfaction Confidence ▪ Concept CONFIDENCE, suppose that support for X → Y is 80%, it means there are 80% chances that X and Y will appear together in a transaction – would be interest to the sales manager ▪ Suppose have another pairs of items (A and B) and support for A → B is 50%. Even thought it is not frequent as X → Y, but whenever A there is 90% chance that B also appears - it would also be a great interest ▪ The probability that a transaction containing the antecedent also contains the consequent. Measures the reliability of the rule. ▪ Confidence and support are important metrics to judge the quality of the association mining rule ▪ Confidence: Confidence ▪ Calculate Confidence of (Bread, Milk) Support and Confidence Lift ▪ A metric used to evaluate the strength of an association between items in a rule. ▪ It helps to determine how much more likely two items are to be bought together compared to being bought independently. ▪ Lift is a useful measure because it normalizes the frequency of co-occurrence, allowing for a more meaningful comparison between different pairs of items, especially when their individual frequencies might differ. ▪ Lift ▪ Calculate lift (Bread -> Milk) Application ▪ Market basket analysis: analyzing consumer purchase patterns to discover associations between products. By identifying which products are often bought together, businesses can design better marketing strategies. ▪ Recommender System: suggest products, services, or content to users based on their preferences or previous behavior. ▪ Customer Segmentation: discovering hidden patterns in customer behavior. This enables businesses to identify customer groups that share similar preferences or purchase habits. ▪ Supply Chain Management : optimize inventory management and supply chain processes by discovering relationships between product demand and supply patterns. ▪ Telecommunication: Churn Prediction - analyze customer behavior and predict churn by identifying patterns that indicate when a customer is likely to leave. ▪ Travel and tourism: Travel agencies and platforms use association rule mining to understand customer travel preferences, and recommend packages, flights, or hotels that are often booked together. THANK YOU 16