Hasil Pencarian  ::  Simpan CSV :: Kembali

"Menggunakan dataset dari salah satu perusahaan cryptocurrency exchange di Indonesia, penelitian ini bertujuan untuk memprediksi churn di cryptocurrency exchange dan menganalisis faktor yang mempengaruhinya. Model yang dikembangkan dalam penelitian ini menggunakan decision tree dan random forest dengan dua kriteria churn yang berbeda. Kriteria churn pertama merupakan kombinasi dari recency dan saldo dalam dompet dan yang kedua hanya menggunakan recency namun lebih personal karena memperhitungkan riwayat jarak antar 2 transaksi dari masing-masing pengguna. Pada kiteria churn pertama, metode undersampling diterapkan sebelum pemodelan karena proporsi churn dan non-churn yang tidak seimbang. Hasilnya model yang dihasilkan dari data undersampling memiliki performa yang terbaik pada model decision tree maupun random forest dengan nilai AUC masing-masing sebesar 0,777 dan 0,787. Hasil dari kedua model juga menunjukkan bahwa penggunaan Google Authenticator dan frekuensi transaksi cryptocurrency merupakan faktor penting untuk menentukan apakah pelanggan akan mengalami churn.

---

Using datasets from one cryptocurrency exchange company in Indonesia, this study aims to predict churn in cryptocurrency exchange and analyze the factor that impacts it. The model developed in this work uses a decision tree and a random forest with 2 different churn criteria. The first criteria is combined the recency and wallet balance amount and the second is used personalized recency (calculate the days between 2 transactions). For the first criteria, the undersampling method is applied before modeling due to imbalanced data. As the result, models from the undersampling dataset have the best performance for the decision tree and the random forest with AUC value of 0,777 and 0,787. Results from both models suggested that the use of Google Authenticator and the frequency of cryptocurrency transactions are important factors to determine whether a customer will experience churn."

"Association rules are used to predict frequent web user behaviors from web usage data. These rules are formed using frequent items. The number of association rules increases as the number of frequent items increases and produces several redundant and non-informative rules. In this paper, five interestingness measures, including cosine, lift, leverage, confidence, and conviction with a constant value of support are compared based on the number of redundant and non-informative rules that they produce. Redundant and non-informative rules are a subset of rules present in the top generated rules. The experimental results suggested that leverage produced the least number of redundant rules in the top rules but also produced the least informative rules among all measures. Lift showed the highest number of redundant rules but the most informative rules among all the measures."