Segmentasi Dan Peramalan Pasar Retail Menggunakan Xgboost Dan Principal Component Analysis

Abstract

The growth of the online retail market in Indonesia is an excellent business opportunity. It is predicted that this growth will continue to move upward due to the increasing internet penetration. With greater exposure to brands, products and offerings, consumers become smarter and wiser in their purchasing decisions. Offering goods and services that match the tastes and behavior of consumers is very important to maintain business continuity. So far, the models developed are divided into two major parts, namely the time series approach and machine learning. In this study, segmentation and forecasting of online retail sector sales were carried out using extreme gradient boosting (XGBoost). The data used in this study is an online retail dataset obtained from the UCI repository. The k-means clustering (KMC) method is applied to determine the target or data class. Principal component analysis (PCA) is applied to reduce data dimensions by eliminating irrelevant features. Model evaluation is based on a confusion matrix and macro average ROC curve. Based on the research results, XGBoost can perform retail data classification well, this can be seen through confusion matrix metrics and ROC curves.

Comparison of Support Vector Machine and XGBSVM in Analyzing Public Opinion on Covid-19 Vaccination

“…To improve this accuracy, this research combines it with the XGBoost algorithm. XGBoost is a decision tree or regression tree-based boosting algorithm [14]. Previous research [15] performed a hybrid SVM with XGBoost and named it XGBSVM.…”

Application of Classification Algorithm for Sales Prediction

“…Offers tailored to local tastes and buying behavior are critical to maintaining business continuity. In this field, intelligent systems and data science can help solve market and consumer segmentation problems [2].…”