Abstract:
Drug purchase transaction data are very valuable treasure in sales. Drug purchase
transaction data are used to generate new knowledge in the transaction database. Drug
purchase transaction data can be managed using the data mining association rules
method. The aim of the study was to apply maximal frequent patterns algorithms to
analyze a patient's drug purchase patterns, look for drug combinations purchased
simultaneously by patients, apply the fp-max association rules algorithm to get rules by
testing the minimum support and minimum confidence desired, and knowing what
variables can affect the association. Drug purchase transaction data were processed by
the data mining method association rule technique using the fp-max algorithm. Drug
purchase transaction data were tested as much as 1909 data with a minimum support of
1% and minimum confidence of 20% which resulted in 8 rules with a lift ratio value of 1
as many as 2 rules. While testing using a minimum support of 2% and so on does not
produce rules because no frequent itemset were produced. Testing lift ratio on 8 rules
concluded no value is worth more than 1, meaning there were no rules that showed a
strong dependence between antacedent and consequence so it can not be used as a
recommendation or prediction of the emergence of a drug due to the emergence of other
drugs. The amount of support count value affected the number of frequent items formed
and the high minimum confidence limit affects the rules produced
