@article{kitko2022identifying,
  title = {Identifying Cannabis Dispensary Purchase Patterns with Market Basket Analysis},
  author = {Paul Kitko},
  journal = {Cannabis Data Science},
  volume = {1},
  number = {1},
  pages = {1--24},
  year = {2022},
  publisher = {Cannlytics},
  url = {\url{https://cannlytics.com/whitepapers}},
  keywords = {Cannabis retail, Market basket analysis},
  abstract = {Market Basket Analysis (MBA) sometimes referred to as
Association Rule Mining, Affinity Analysis or Frequent Itemset Mining,
was developed as a method to evaluate \"if / then\" associations that
arise between elements in a dataset. Historically MBA rule sets have
been applied to retail grocery stores' Point of Sale data to develop
likely product associations that can then be used to anticipate and
recommend combinations of future purchases. These recommendations or
\"cross--sells\" have been found to be useful in improving retail sales
volume. The newly legalized recreational cannabis market offers an
opportunity to apply MBA to an unexplored retail industry. This
project used MBA on a retail cannabis dataset representing multiple
dispensaries across the state of Washington. The project's purpose
was to verify if MBA was feasible in uncovering useful product
association rules from a cannabis sales dataset to use in cross-selling
recommendations. The results of the study show that it is possible to
derive meaning MBA rule sets from cannabis retail data but that some
limitations were uncovered that offer three future opportunities for
research. First that similar product with highly differentiated names
may need to be re-categorized into more generalized and meaningful
products. Second, that it is possible that product churn may introduce
signal noise into the MBA process resulting in a higher number of less
useful rule sets. Third, that cannabis customers tend to purchase
within product families which is an atypical finding in MBA and
should be further explored.}
}