Klasifikasi Kategori Produk untuk Manajemen Keuangan Remaja menggunakan Algoritma Long Short-Term Memory

Abstract

Generation Z often faces difficulties in managing their finances due to impulsive spending habits and a lack of financial planning, which can lead to long-term issues such as overspending and minimal savings. This research aims to develop a category classification model that can be integrated into a financial tracking application to help young people manage their money more effectively. The main feature of the application is an automated system that classifies product names into expense categories such as food, transportation, and shopping using a Long Short-Term Memory (LSTM) model. LSTM was chosen for its ability to understand word sequences and text context, which is essential in product grouping. The dataset used consists of 4,499 product entries divided into three categories: 1,488 for food, 1,682 for transportation, and 1,329 for shopping. The model was trained using a supervised learning approach, with data split for training and testing. The model achieved 86% accuracy on both validation and test data, with additional metrics such as precision, recall, and F1-score indicating good performance. This study contributes by applying innovative preprocessing techniques and oversampling to address data imbalance, which is expected to enhance the model's accuracy in classifying expenses.