MIS 301 Lecture Notes - Lecture 20: Customer Experience, Randomized Experiment, Affinity Analysis

• School: UT-Austin
• Department: Management Information Systems
• Course: MIS 301
• Professor: Professor Choi

Document Summary

Mis 301- lecture 3. 4 data mining and amazon. Ex: at a supermarket: milk and cereal. Support count: number of times that a certain product has been bought. Support: probability that a certain product is bought. Support = support count / number of transactions. Support of a product shows its impact in terms of the overall size: association rule: (antecedent) (consequent) Confidence level: the rate at which consequents will be found given that the antecedent has occurred. P(consequent | antecedent) = p (antecedent and consequent) / In this case: 2/5 / 3/5 = 2/3 is the confidence level (confidence level slide p(beer | milk&diaper)) Lift ratio: indicator that tells if an association rule is useful or not. The support of a product indicates its impact. Confidence tells what rate the consequent will be found given that the antecedent has already been found. Lift ratio tells us if cross-selling strategy is useful.