212 Mart Rambutan Street on Pekanbaru City is a company engaged in retail. Meeting the needs of consumers and making the right decision in determining the sales strategy is a must. One way to find out market conditions is to observe sales transaction data using data mining. The data mining method commonly used to analyze market basket (Market Basket Analysis) is the Association Rule. The Association Rule can provide product recommendations and promotions, so that the marketing strategy is more targeted and the items promoted are the customer's needs. At 212 Mart, the determination of product promotion is obtained from the analysis of sales transaction data reports, which are based on the most sold products and the expiration date. Often the product being promoted does not fit the customer's needs. The purpose of this study is to apply the K-Medoids algorithm for clustering on FP-Growth in producing product recommendation rules on a large number of datasets so that they can provide technical recommendations / new ways to the 212 Mart in determining product promotions. The results obtained are from the experiments the number of clusters 3 to 9 obtained optimal clusters of 3 clusters based on the validity test of the Davies Bouldin Index with a value of 0.678. With a minimum support value of 5% - 9% and a minimum value of 50% confidence, the result is that the Association Rule is found only in cluster 3 with 5 rules.

Diabetes juga dikenal sebagai penyakit metabolik kronis yang dipicu karena pankreas tidak menghasilkan insulin yang cukup. Pada masa kehamilan, tubuh akan meningkatkan hormon-hormon yang juga mempengaruhi kerja insulin. Pada penelitian ini data yang digunakanlah adalah kemungkinan resiko ibu hamil terkena diabetes, data tersebut diolah memakai teknik data mining yaitu naïve bayes. Setelah dilakukan pengolahan data memakai teknik klasifiaksi data mining algorithma  naïve bayes, hasil yang didapat untuk pembagian data menggunakan K-Fold Cross Validation K=10 pada algoritma naïve bayes didapat lah hasil 75,78% dan untuk penggolahan menggunakan knn dengan nilai K=25 didapat hasil 74,48%. Dari hasil tersebut naïve bayes lebih baik dibandingkan K-Nearest Neighbor (KNN).