Penerapan Optimasi Gradient Boosting dalam Prediksi Nilai Transaksi Pelanggan di E-CRM Fluffy Cat Shop

Abstract

The development of e-commerce encourages companies to optimally leverage customer data through an Electronic Customer Relationship Management (E-CRM) system. Fluffy Cat Shop, an online store for cat supplies, faces challenges in accurately predicting customer transaction value. This study aims to optimize the Gradient Boosting algorithm for predicting customer transaction value within the E-CRM system of Fluffy Cat Shop. The research methods include collecting customer transaction data, data preprocessing (cleaning, encoding, and normalization), building a Gradient Boosting model, and optimizing hyperparameters using the Grid Search method. Model evaluation is conducted using the MAE, RMSE, and R² Score metrics. The results show that after optimization, the model’s performance improves with an R² Score of 0.8, indicating that the model can explain 80% of the variation in customer transaction value. The error values also decrease compared to the initial model.