- 學位論文
使用長短期記憶最佳化模型之花卉批發價格預測
Using LSTM Model with Optuna for Predicting Flower Wholesale Prices
本文將於2029/05/31開放下載。若您希望在開放下載時收到通知，可將文章加入收藏。
摘要
本研究旨在使用深度迴歸模型預測台灣批發市場香水百合花的拍賣價格。在現今社會中，香水百合是一種很受歡迎的花卉，同時也是一種高價值的商品。百合花價格的預測主要是由農民依據經驗進行預估，但這樣的方法對於缺乏市場敏感度或剛開始從事農業種植的花農較難以達到良好的預測效果。因此，本研究將應用深度迴歸模型及5年的交易歷史價格建立預測模型，其採用7、14、21及28天的歷史批發價格預測未來1至7天的批發交易價格。研究中採用長短記憶模型(Long Short-Term Memory，LSTM）及Optuna自動超參數搜尋方法建立最佳化的香水百合花批發價格預測模型。研究結果顯示Optuna能提升預測模型的效能，相較未優化及優化的模組後均能有效的提升其效能，同時也發現以21天歷史價格資訊建立預測模型對未來1至7天的平均效能是最佳的。本研究結果將可提供花農進行較為精準的價格預測方法並決定交貨至拍賣市的最佳時機，以獲得更好的利潤。
並列摘要
This study aims to use deep regression models to predict the auction prices of perfumed lilies in Taiwan's wholesale markets. In today's society, perfumed lilies are a popular flower and a high-value commodity. Farmers typically estimate lily prices based on their experience, but this method may not achieve good predictive results for those lacking market sensitivity or new to agricultural cultivation. Therefore, this study will apply deep regression models and use five years of historical transaction prices to establish a predictive model, which employs historical wholesale prices from 7, 14, 21, and 28 days to predict wholesale transaction prices for the next 1 to 7 days. The study adopts the Long Short-Term Memory (LSTM) model and Optuna automatic hyperparameter search method to establish an optimized predictive model for the wholesale prices of perfumed lilies. The results show that Optuna can enhance the model's performance, significantly improving it compared to non-optimized models, and the best average performance for predicting the next 1 to 7 days is achieved using a 21-day historical price dataset. The results of this study can provide flower farmers with a more accurate method for price prediction and help determine the best timing for delivering their products to the auction market, thereby obtaining better profits.
參考文獻
參考文獻
1. 農糧署. (2022). 中華民國111年臺灣地區農產品批發市場年報. https://amis.afa.gov.tw/doc/110年報.pdf
2. Ajik, E. D., Obunadike, G. N., & Echobu, F. O. (2023). Fake News Detection Using Optimized CNN and LSTM Techniques. Journal of Information Systems and Informatics, 5(3), 1044–1057. https://doi.org/10.51519/journalisi.v5i3.548
3. Akiba, T., Sano, S., Yanase, T., Ohta, T., & Koyama, M. (2019). Optuna: A Next-generation Hyperparameter Optimization Framework. Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2623–2631. https://doi.org/10.1145/3292500.3330701
4. Alizadeh, B., Ghaderi Bafti, A., Kamangir, H., Zhang, Y., Wright, D. B., & Franz, K. J. (2021). A novel attention-based LSTM cell post-processor coupled with bayesian optimization for streamflow prediction. Journal of Hydrology, 601, 126526. https://doi.org/10.1016/j.jhydrol.2021.126526