以衛星資訊建立預測玉米產量之模型;Predicting maize yields with satellite information

NCU Institutional Repository > 管理學院 > 企業管理研究所 > 博碩士論文 > Item 987654321/88463

請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/88463

題名:	以衛星資訊建立預測玉米產量之模型;Predicting maize yields with satellite information
作者:	施又升 Shih, You-Sheng
貢獻者:	企業管理學系
關鍵詞:	衛星遙測;遙測面積;植被指數;迴歸分析;Satellite telemetry;Telemetry area;Vegetation Index;Regression Analysis
日期:	2022-06-30
上傳時間:	2022-07-14 10:16:00 (UTC+8)
出版者:	國立中央大學
摘要:	美國的玉米是全球產量最大的作物，故美國的玉米產量足以牽動整個穀物市場的價格。大多數預測美國玉米產量的研究聚焦於植被指數對產量的影響，而本研究根據植被覆蓋率越高且種植面積越大則作物產量越大的假設，將針對植被指數和遙測面積建立迴歸模型預測美國玉米產量。使用的迴歸模型有多元線性迴歸、偏最小二乘迴歸、逐步迴歸以及利用高斯核的支持向量迴歸，最後實驗結果以高斯核(Radial basis function kernel)的支持向量迴歸表現最佳，??2值為 0.94。;Unite States of America harvests the largest crop of maize in the world. The volume it grows, therefore, critically affects many countries and industries. Predicting the yields thus have discussed by prior studies. Recently, with the conveniently available of Satellite images, several research has attempted to make prediction vegetation index on yield. However, this research argues that besides vegetation index, the data of telemetry area are also needed, as higher vegetation coverage and larger planting area lead to greater crop yields. This research therefore, strives to derive 9 years of data for 4 most important states to train various regression models, which include multivariable linear regression, partial least squares regression, stepwise regression, and support vector regression with Gaussian kernel. The result shows that the support vector regression with Gaussian kernel (Radial basis function kernel) performed the best, with R^2 value reaches 0.94.
顯示於類別:	[企業管理研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	61	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....