Revenue Forecasting Models from Point-of-Sale Data for Electronic Payment and Money Institutions
4th Cognitive Models and Artificial Intelligence Conference, AICCONF 2026, Prague, Çek Cumhuriyeti, 24 - 25 Nisan 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/aicconf69182.2026.11600687
- Basıldığı Şehir: Prague
- Basıldığı Ülke: Çek Cumhuriyeti
- Anahtar Kelimeler: Electronic Payment and Money Institutions, Machine Learning, Revenue Forecasting
- Çukurova Üniversitesi Adresli: Evet
Özet
Electronic payment and money institutions offer innovative solutions in the financial services sector by digitizing financial transactions. These institutions generate revenue through various products and services. Point of Sale (POS) is one of these products. Revenue forecasting for POS is a critical process that directly affects financial efficiency and market competitiveness. In this study, POS revenue forecasting models have been developed using Convolutional Long-Short Term Memory (ConvLSTM), Convolutional Neural Network (CNN-LSTM), Seasonal Auto Regressive Integrated Moving Average (SARIMA), Categorical Boosting (CatBoost), Support Vector Machine (SVM) and ensemble learning. The dataset has been created using data obtained from Moka United. Monthly and weekly forecasting models have been developed for the months of September and December 2024. The effects of F-Regression feature selection and sequential/periodic lookback integration on the performance of the forecasting models have been analyzed. The performance of the developed forecasting models has been evaluated using Mean Absolute Percentage Error (MAPE). As a result, the CNN-LSTM model achieved the lowest MAPE value.