Analisis Komparatif Pola Pembelian Konsumen Marketplace pada Periode Weekdays dan Weekends Menggunakan Algoritma FP-Growth

Product purchasing patterns on e-commerce platforms can be studied through transaction times, such as differentiating between weekdays and weekends. This study aims to compare product association patterns in the household goods category during weekdays and weekends using the FP-Growth algorithm, while evaluating their recommendation performance using Top-K metrics. Knowledge Discovery in Databases (KDD)-based processing was applied to e-commerce transaction data, yielding 17,039 clean records from an initial dataset of 20,848 raw data, consisting of 12,898 weekday transactions (75.7%) and 4,141 weekend transactions (24.3%). Association patterns were extracted using the FP-Growth algorithm, and recommendation performance was evaluated at K = 1, 3, 5. The results indicate differences in product characteristics; weekdays are dominated by combinations of daily necessity products (maximum Lift 4.1226), whereas weekends exhibit a wider variety of combinations (maximum Lift 4.2135). Testing identifies K = 3 as the optimal parameter, with F1-Score@3 values of 0.0445 for weekdays and 0.0814 for weekends. Findings from the implementation of FP-Growth and Top-K present association structures and objective recommendation performance, serving as a technical reference in implementing product recommendation system strategies.