Analisis Pola Kombinasi Pembelian Obat pada Apotek Menggunakan Perbandingan Metode Asosiasi Apriori dan FP-Growth

Anaya, Agni Ilmi (2026) Analisis Pola Kombinasi Pembelian Obat pada Apotek Menggunakan Perbandingan Metode Asosiasi Apriori dan FP-Growth. Skripsi thesis, Universitas Islam Imam Bonjol Padang.

	Text (Cover - Daftar Isi) Cover - Daftar Isi.pdf - Published Version Download (1MB)
	Text (BAB I) BAB I.pdf - Published Version Download (282kB)
	Text (BAB III) BAB III.pdf - Published Version Download (400kB)
	Text (BAB V - Daftar Pustaka) BAB V - Daftar Pustaka.pdf - Published Version Download (291kB)
	Text (Full Text) SKRIPSI FULL TEXT.pdf - Published Version Restricted to Repository staff only Download (4MB)

Abstract

Pharmacy sales transactions occur daily with high product variation, forming purchase combinations that can be leveraged to arrange product layout, control inventory, and design sales strategies. However, this potential has not been optimally utilized by small to medium-scale pharmacies that still record transactions manually, such as Apotek Gretta Solok, leaving the data unready for analysis. Previous studies comparing the Apriori and FP-Growth algorithms have generally used large, well-structured, digitized datasets, so their results may not reflect the performance of both algorithms on limited pharmacy data derived from manual record-keeping. Based on this problem, this study aims to produce valid transaction data, identify drug purchase combinations using the Apriori and FP-Growth algorithms, and compare the performance of both algorithms based on the parameters of support, confidence, number of association rules, and processing time. This study used a quantitative comparative approach with the CRISP-DM framework, involving digitization of Apotek Gretta Solok's manual transaction data for the period of October 2025-February 2026 using Claude's multimodal model, verification, drug name standardization, and transformation into basket format, resulting in 4,202 valid drug records from approximately 5,599 initial records. The model was tested on five minimum support scenarios (10%, 20%, 30%, 40%, 50%) with a fixed minimum confidence of 40%. The results show that Apriori and FP-Growth produced identical numbers and content of association rules across all scenarios, with scenario S3 (30% support, 40% confidence) as the best parameter, yielding 18 valid rules (lift greater than 1). The Cetirizine-Paramex combination had the highest support (41.6%), forming two main purchase pattern groups: fever/pain relievers and allergy/flu medications. FP-Growth was slightly more time-efficient and was selected as the recommended final model. This study concludes that imperfect small-scale pharmacy transaction data can still be processed into an analysis-ready dataset, and the resulting association patterns can serve as a basis for shelf arrangement and stock management recommendations for Apotek Gretta Solok.

Actions (login required)

View Item