Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/92917
標題: Market Trading Volume Predict System: Case of Pomelo in Taiwan
台灣文旦市場到貨量預估系統
作者: 黃仁佑
Jen-Yu Huang
關鍵字: Data mining
genetic
algorithm
pomelo
predict
文旦柚
市場到貨量
資料探勘
基因演算法
引用: 1. 行政院農業委員會102年9月農業產銷概況,102年(第247期-258期) 2. 許雅筑。 2012。 '產銷問題 政府應積極作為.' 。喀報第一二七期。 3. 洪忠修。2003。我國農牧業農情調查之檢討與展望,農政與農情 135期。 4. 張汶肇。2013。雲林、嘉義及臺南地區文旦柚產業概況,農友月刊。18-20。 5. 林芳存。1994。麻豆文旦果實生育變化與品質之研究。臺灣大學園藝學研究所碩士論文。 6. 張淳淳、張汶肇、卓家榮、林明瑩。2014。臺南區麻豆文旦健康管理生產模式之建立。102年度重點作物健康管理生產體系及關鍵技術之研發成果研討會論文集。58-69。 7. 行政院農業委員會。田邊好幫手農業生產預測。網址: http://m.coa.gov.tw/outside/forecast/Search.aspx 8. 陳溪潭。1996。麻豆文旦春梢生長與結果習性之探討。中國園藝42(1):78- 88。 9. 行政院農業委員會。2006。農產品交易行情站。網址: http://amis.afa.gov.tw/。 10. 農糧署統計室。2011。農產品產地價格查報系統。網址: http://apis.afa.gov.tw/pagepub/AppInquiryPage.aspx 11. 交通部中央氣象局。網址:http://www.cwb.gov.tw/V7/climate/agri/agrb.htm 12. 莊浚釗、李宗翰。2011。文旦之土壤和肥培管理技術研究 。桃園區農業改良場研究彙報 69:59-73,。 13. Cabrera, V. E., S. S. Jagtap, and P. E. Hildebrand. 2007. Strategies to limit (minimize) nitrogen leaching on dairy farms driven by seasonalclimate forecasts. Agriculture, Ecosystem and Environment 122: 479-489. 14. Brown, C., P. Rogers, and U. Lall. 2006. Demand Management of Groundwater with Monsoon forecasting. Agricultural Systems 90: 293-311. 15. Canesio, P., C. Jason, H. Peter, M. John, and P. Kevin. 2007. Profitable use of seasonal climate forecasts: Farm management case studies in the Philippinnes and Australia. Philippine J. Crop Science 32: 41-42 16. Wen, Qian, et al. 'Daily Sales Forecasting for Grapes by Support Vector Machine.' Computer and Computing Technologies in Agriculture VII. Springer Berlin Heidelberg, 2014. 351-360. 17. David Freedman; Robert Pisani, Roger Purves. Statistics. Norton & Company. 1998: 148. 18. Kusiak, Andrew, et al. 'Modeling and prediction of rainfall using radar reflectivity data: A data-mining approach.' Geoscience and Remote Sensing, IEEE Transactions on 51.4 (2013): 2337-2342. 19. Y.S. Yun, nHybrid Genetic Algorithm With Adaptive Local Search Scheme,S Computers & Industrial Engineering,VOL.51,NO.1,2006.128-141. 20. Mühlenbein, Heinz, and Dirk Schlierkamp-Voosen. 'Predictive models for the breeder genetic algorithm i. continuous parameter optimization.' Evolutionary computation 1.1 (1993): 25-49. 21. Mitchell, Melanie. An introduction to genetic algorithms. MIT press, 1998.
摘要: In Taiwan, pomelo is an occasional fruit of Mid-Autumn Festival and the annual demand is concentrated at this time, prone to supply and demand imbalance. If the amount of pomelo production can be accurately predicted, according to the prediction, not only the farmers can adjust their planting plan, but the government can also adjust countermeasures to stablize market price. Pemelo production is very vulnerable to soil quality, cultivation techniques, pest, and climate factors during its growth period. These factors will determine the amount of Pomelo production; hence, accurately predicting the future pomelo production is too complicated and impractical. The traditional methods adopt estimated acreages and the amount of production per unit area to calculate the future amount of pomelo production. However, it cannot offer an accurate prediction result as a result of many factors about pomelo production. Since the actual pomelo production amount cannot be acquired, so the intelligent information technology cannot be effectively applied to predict pomelo production amount. In Taiwan, the trading volume in wholesale market is about 77.21% of the total pomelo production. The trading volume in wholesale markets and production is highly co-related. Therefore it is feasible to predict the pomelo trading volume in wholesale markets and then to derive pomelo production from the predicted trading volume. In the wholesale market, each trading volume is precisely recorded and there are only a few wholesale markets in Taiwan relative to the pomelo planting areas. It is more economic, cost-saving, and accurate to predict the pomelo trading volume in wholesale markets. In this research , a genetic algorithm based prediction model about pomelo trading volume of a wholesale market is provided. This model is employed to predict the pomelo trading volume of wholesale markets in Madou of Tainan and Chiayi. The experimental results demonstrate that the model provides only 3.5% error prediction rate in Madou, and 3.2% in Chiayi.
文旦為台灣中秋節時的應景水果,每年的需求量集中在此時,因此易產生供需失調,容易造成農民損失。若能對文旦產量進行精準預測,不僅能提供農人對種植面積進行調節,也能提供政府制定調節市場供應量因應對策之參考,期能平穩交易價格,避免失衡情況發生。文旦產量易受到種植土壤、棵樹、栽培技術、病蟲害、氣候,與採收時間等因素之影響。欲精準預測文旦未來產量實太複雜而不可行。現有系統借估算種植面積與單位面積產量來推算產量,但因無法取得歷年各地區文旦實際產量資料,智慧型資訊技術也難以提供精準預測。 台灣批發市場貨物的流通量約為總生產量的77.21%,因有專門人員進行記錄,故批發市場到貨量能有完整的紀錄;全國批發市場僅分布於少數地點,資料取得也較為容易;此外產量與到貨量彼此間也有高度共相關。因此透過預測各個批發市場的文旦到貨量後,再將其換算為產量實為可行;此不僅能獲得較為精確的預測值,也能省下大量預測的成本。故本研究將從已估算之種植面積與種植期間的氣候因素,來預測文旦的未來市場到貨量。本研究分別建立了台南麻豆市場與嘉義果菜市場兩地區的文旦到貨量預測模型。實驗結果顯示台南地區整體的預測到貨量與實際到貨量的誤差率約為3.5%,嘉義地區則約為3.2%,驗證了此模型能精確的預測市場到貨量。本研究也從此模型,引導出在每個不同生長時段裡,影響文旦生產的不同重要因素。此能夠提供給農民在栽植時,對每個不同時段裡,須注意哪些重要生長因素。
URI: http://hdl.handle.net/11455/92917
其他識別: U0005-1208201511113400
文章公開時間: 10000-01-01
Appears in Collections:資訊管理學系

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.