The use of artificial intelligence methods to predict income and expense trends from semi-unstructured data for accounting planning for pineapple farmers in Phitsanulok province

Abstract

 This study explores the use of artificial intelligence (AI) techniques, particularly Natural Language Processing (NLP), to predict income and expense trends for pineapple farmers in Ban Yang Subdistrict, Nakhon Thai District, Phitsanulok Province. The research focuses on extracting and analyzing semi-unstructured accounting data, primarily from PDF files, provided by 30 farmers with prior accounting experience. The data was processed using a custom NLP algorithm to classify financial records into payments and income categories. Time series forecasting models—Prophet, LSTM, and ARIMA—were applied to predict future trends. The results revealed significant differences in model performance, with ARIMA outperforming both LSTM and Prophet. ARIMA exhibited the lowest Mean Absolute Error (MAE) of 34.1084 and the highest R-squared (R²) of 0.9901, indicating superior prediction accuracy and explanatory power. LSTM showed a MAE of 71.0920 and an R² of 0.9511, demonstrating good accuracy but with higher MAE and lower R² than ARIMA. Prophet demonstrated the highest MAE of 603.8044 and the lowest R² of -3.2273, reflecting poor performance. Based on these findings, ARIMA is identified as the most suitable model for this dataset, with potential for further optimization and exploration of additional features to improve predictive accuracy.