RAMADHAN, MUHAMMAD RIZKI (2026) ANALISIS KOMPARATIF MODEL XGBOOST DAN RANDOM FOREST DALAM PREDIKSI REVENUE E- COMMERCE BERBASIS TIME SERIES. S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
Cover.pdf Download (415kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
Bab 1.pdf Restricted to Registered users only Download (34kB) |
|
|
Text (BAB II)
Bab 2.pdf Restricted to Registered users only Download (187kB) |
|
|
Text (BAB III)
Bab 3.pdf Restricted to Registered users only Download (153kB) |
|
|
Text (BAB IV)
Bab 4.pdf Restricted to Registered users only Download (192kB) |
|
|
Text (BAB V)
Bab 5.pdf Restricted to Registered users only Download (32kB) |
|
|
Text (DAFTAR PUSTAKA)
Daftar Pustaka.pdf Restricted to Registered users only Download (142kB) |
|
|
Text (LAMPIRAN)
Lampiran.pdf Restricted to Registered users only Download (541kB) |
Abstract
The rapid growth of e-commerce has generated large volumes of transaction data that can serve as a valuable basis for business decision-making, particularly for revenue forecasting. This study compares the performance of XGBoost and Random Forest algorithms in predicting daily revenue using time series data. Transaction data from an e-commerce platform spanning 2020–2024, originally recorded at the item level, was first transformed into order-level data before undergoing preprocessing, daily aggregation, and feature engineering, resulting in 1,245 daily observations with 13 predictor variables and one target variable, Total Revenue. The dataset was split using a time series split approach, allocating 80% for training and 20% for testing, while model performance was evaluated using three metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the Coefficient of Determination (R²). The results show that Random Forest outperformed XGBoost, achieving an MAE of IDR 283,955.27, an RMSE of IDR 457,777.99, and an R² of 0.8871, compared to XGBoost's MAE of IDR 362,813.31, RMSE of IDR 529,629.85, and R² of 0.8488. This performance gap suggests that the daily revenue data in this study is better captured by a bagging-based ensemble approach such as Random Forest, which tends to be more robust to fluctuations and noise than a boosting-based approach like XGBoost. These findings indicate that Random Forest can be recommended as a more reliable model for forecasting daily revenue on similar e-commerce datasets. Kata kunci: Revenue Prediction, E-Commerce, Random Forest, XGBoost, Time Series, Machine Learning. Pertumbuhan e-commerce yang pesat turut menghasilkan volume data transaksi yang besar, dan data tersebut berpotensi diolah menjadi dasar pengambilan keputusan bisnis, salah satunya untuk memprediksi revenue. Penelitian ini membandingkan kinerja algoritma XGBoost dan Random Forest dalam memprediksi revenue harian berbasis data time series. Data transaksi e-commerce periode 2020–2024 yang awalnya berada pada level item terlebih dahulu ditransformasikan ke level order, kemudian melalui tahap preprocessing, agregasi harian, dan feature engineering hingga menghasilkan 1.245 observasi dengan 13 variabel prediktor dan satu variabel target, yaitu Total_Revenue. Pembagian data dilakukan dengan pendekatan time series split (80% data latih dan 20% data uji), sementara evaluasi model menggunakan tiga metrik, yaitu MAE, RMSE, dan R². Hasil pengujian menunjukkan Random Forest memberikan kinerja yang lebih baik dibandingkan XGBoost, dengan nilai MAE Rp283.955,27, RMSE Rp457.777,99, dan R² 0,8871, sedangkan XGBoost menghasilkan MAE Rp362.813,31, RMSE Rp529.629,85, dan R² 0,8488. Selisih performa ini mengindikasikan bahwa karakteristik data revenue harian pada penelitian ini lebih cocok ditangani oleh pendekatan ensemble bagging seperti Random Forest, yang cenderung lebih stabil terhadap fluktuasi dan noise dibandingkan pendekatan boosting seperti XGBoost. Dengan demikian, Random Forest dapat direkomendasikan sebagai model yang lebih andal untuk memprediksi revenue harian pada dataset e-commerce sejenis. Kata kunci: Prediksi Revenue, E-Commerce, Random Forest, XGBoost, Time Series, Machine Learning.
Actions (login required)
 |
View Item |
