Rekomendasi Pemilihan Musik Menggunakan Market Basket Analysis (MBA) dengan Algoritma Apriori, Frequent-Pattern Growth (FP-Growth), Equivalence Class Transformation (ECLAT)

Sumarsono, Abraham Muhammad Prayitno (2024) Rekomendasi Pemilihan Musik Menggunakan Market Basket Analysis (MBA) dengan Algoritma Apriori, Frequent-Pattern Growth (FP-Growth), Equivalence Class Transformation (ECLAT). Other thesis, Institut Tekonologi Sepuluh Nopember.

Text 5003201099_Undergraduate-Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk mendapatkan rekomendasi pada lagu dan genre. Rekomendasi didapatkan dengan menggunakna konsep similarity. Salah satu penerapan konsep similarity ini adalah Market Basket Analysis (MBA) yang biasanya digunakan dalam bidang retail. Metode MBA ini dilakukan dengan beberapa tahapan yaitu preprocessing (cleaning data, transformasi data biner, menghitung nilai support, menentukan nilai minimum support, dan mereduksi data dengan minimum support sebagai batas), melakukan pencarian frequent itemset, menghitung nilai confidence, menentukan nilai minimum confidence, membuat rekomendasi, dan menghitung tingkat akurasi algoritma. Tahapan tersebut dilakukan terhadap dua data utama yaitu untuk data lagu dan genre. Pada penelitian ini, digunakan tiga algoritma yang umum digunakan saat melakukan analisis dengan menggunakan metode MBA yaitu, algoritma apriori, frequent pattern growth, dan equivalence class transformation. Melalui penelitian ini, diketahui bahwa ketiga algoritma menghasilkan jumlah frequent itemsets yang sama untuk ketiga algoritma, tetapi jumlah rules yang dihasilkan berbeda untuk genre. Berdasarkan ketepatan rekomendasi, ketiga algoritma menghasilkan rata-rata akurasi sebesar 32,23% untuk lagu. Sedangkan untuk data genre, dihasilkan rata-rata akurasi sebesar 95,27% untuk algoritma apriori dan FP-Growth serta 93,48% untuk algoritma ECLAT. Rata-rata akurasi pada genre memiliki nilai yang lebih tinggi karena rekomendasi judul lagu lebih spesifik dibandingkan dengan genre yang memiliki cakupan lebih luas. Sehingga, dikatakan bahwa ketiga algoritma direkomendasikan untuk rekomendasi pemilihan musik. Namun, algoritma equivalence class transformation memakan waktu lebih lama untuk melakukan perhitungan sehingga tidak efisien secara waktu. Rekomendasi genre memiliki akurasi lebih tinggi dibanding rekomendasi lagu.
=================================================================================================================================
This research aims to get recommendations for songs and genres. Recommendations are obtained using the similarity concept. One application of this similarity concept is Market Basket Analysis (MBA) which is usually used in the retail sector. This MBA method is carried out in several stages, namely preprocessing (cleaning data, binary data transformation, calculating support values, determining minimum support values, and reducing data with minimum support as a limit), searching for frequent itemsets, calculating confidence values, determining minimum confidence values, make recommendations, and calculate the accuracy level of the algorithm. This stage is carried out on two main data, namely song and genre data. In this research, three algorithms are used that are commonly used when conducting analysis using the MBA method, namely, the a priori algorithm, frequent pattern growth, and equivalence class transformation. Through this research, it is known that the three algorithms produce the same number of frequent itemsets for the three algorithms, but the number of rules produced is different for genre. Based on recommendation accuracy, the three algorithms produce an average accuracy of 32.23% for songs. Meanwhile, for genre data, an average accuracy of 95.27% was obtained for the apriori and FP-Growth algorithms and 93.48% for the ECLAT algorithm. The average accuracy in the genre has a higher value because the song title recommendations are more specific compared to genres which have a wider scope. So, it is said that the three algorithms are recommended for music selection recommendations. However, the equivalence class transformation algorithm takes longer to perform calculations so it is not time efficient. Genre recommendations have higher accuracy than song recommendations.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Sistem Rekomendasi, Market Basket Analysis, Lagu, Recommendation System, Song
Subjects:	Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Abraham Muhammad Prayitno Sumarsono
Date Deposited:	29 Aug 2024 02:34
Last Modified:	29 Aug 2024 02:34
URI:	http://repository.its.ac.id/id/eprint/115557

Actions (login required)

View Item