Prediksi Penjualan untuk Rekomendasi Paket Produk sebagai Strategi Pemasaran Menggunakan XGBoost dan Random Forest

Sabrina, Dayini Nur (2025) Prediksi Penjualan untuk Rekomendasi Paket Produk sebagai Strategi Pemasaran Menggunakan XGBoost dan Random Forest. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) COVER.pdf Download (1MB)
	Text (BAB 1) BAB 1.pdf Download (187kB)
	Text (BAB 2) BAB 2.pdf Restricted to Repository staff only until 17 June 2027. Download (474kB)
	Text (BAB 3) BAB 3.pdf Restricted to Repository staff only until 17 June 2027. Download (482kB)
	Text (BAB 4) BAB 4.pdf Restricted to Repository staff only until 17 June 2027. Download (794kB)
	Text (BAB 5) BAB 5.pdf Download (175kB)
	Text (Daftar Pustaka) DAFTAR PUSTAKA.pdf Download (152kB)
	Text (Lampiran) LAMPIRAN.pdf Restricted to Repository staff only Download (257kB)

Abstract

This study aims to build a sales prediction model based on sales transaction data by implementing the XGBoost and Random Forest algorithms in a hybrid manner through an ensemble learning approach. The study was conducted at the Infinity Jar online store to predict the number of product sales and provide marketing strategy recommendations in the form of product package combinations based on demand categories. The research method includes data preprocessing, modeling, hyperparameter tuning, model evaluation using Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) metrics, and visualization of results. The results show that the hybrid model with a bagging approach, which combines the prediction results of XGBoost and Random Forest using Voting Regressor, produces the best performance with an MSE value of 5.36, MAE of 1.00, and MAPE of 14.39%. The MSE value describes the average squared error, MAE shows the average absolute difference between the actual and predicted values, while MAPE represents the average percentage error to the actual value. The smaller the values of these three metrics, the better the model performance. The hybrid model proved superior to the single model, where Random Forest produced an MSE of 5.45, MAE of 1.09, and MAPE of 23.48%, and XGBoost with an MSE of 6.91, MAE of 1.27, and MAPE of 33.01%. In addition, this research produced 18 product package combinations as marketing strategy recommendations based on the classification of products with high and low demand. These recommendations are expected to help optimize stock management and increase sales. Thus, the combination of XGBoost and Random Forest algorithms proved effective in modeling sales predictions and supporting data-based marketing strategy decision making. Keywords: Sales Prediction, Marketing Strategy, XGBoost, Random Forest, Ensemble Learning, Voting Regressor.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Rahmat, Basuki	NIDN0023076907	basukirahmat.if@upnjatim.ac.id
Thesis advisor	Puspaningrum, Eva Yulia	NIDN0005078908	evapuspaningrum.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Dayini Nur Sabrina

Date Deposited:

19 Jun 2025 04:21

Last Modified:

19 Jun 2025 04:21

URI:

https://repository.upnjatim.ac.id/id/eprint/38585