查詢結果分析
相關文獻
- 群集檢測在考慮檢測門檻下信賴區間之比較--以農業之兩組資料為例
- Confidence Intervals for Multivariate Process Capability Indices Using Bootstrap
- 玻璃證物化學分析數據之自動化處理
- 從實務工作者觀點分析成人學習的動機--以復興工商專校附設專科進修學校為例
- 科學證據與侵權行為法:美國有關邊得克汀訴訟的省思
- 兩個常態母體中SN比之差的廣義信賴區間
- 什麼是信賴區間(Confidence Interval)﹖
- 以核估式估計中位有效劑量之研究
- 以Delta法求中位有效劑量及相對效能的信賴區間
- Hybrid Resampling Methods for Confidence Intervals
頁籤選單縮合
題 名 | 群集檢測在考慮檢測門檻下信賴區間之比較--以農業之兩組資料為例=Comparison of Confidence Interval for Proportions Estimated by Using Group Testing under the Existence of a Threshold of Detection--Two Cases in Agricultural Science as Examples |
---|---|
作 者 | 鐘智瑋; 鄧汀欽; 蔣國司; | 書刊名 | 作物、環境與生物資訊 |
卷 期 | 11:3 2014.09[民103.09] |
頁 次 | 頁129-144 |
分類號 | 434.28 |
關鍵詞 | 群集檢測; 檢測門檻; 信賴區間; 覆蓋機率; 基因改造生物; Group testing; Threshold of detection; Confidence interval; Coverage probability; Genetically modified organism; |
語 文 | 中文(Chinese) |
中文摘要 | 在農業上群集檢測為一可降低成本與提高效率之檢測方法,此一統計方法可應用於陽性反應個體的估計,其檢測流程如下,首先將欲檢測的個體劃分為數個群集,再分別針對各個群集進行檢測,若檢測結果為陽性反應,則稱該群集內至少有一陽性之檢測單位;若檢測結果為陰性,則稱該群集內之所有檢測單位皆為陰性,此方法特別適用於陽性反應個體數較少的情況;若當檢測方法本身存在有檢測門檻,隨著群集內檢測單位數的增加,可能造成一群集被檢測為陽性的機率遠低於其門檻值,進而導致偽陰性的檢測結果。本研究主要在探討當使用群集檢測且帶有檢測門檻時,群集內檢測單位為陽性之機率估計問題,提出並比較四種信賴區間方法(Wald、LikelihoodRatio、Score及Exact),目的在探討此四種區間方法在群集檢測帶有檢測門檻時表現之優劣處;其中信賴區間之評估準則,以覆蓋機率、期望寬度及平均絕對偏差為標準,進而歸納出較佳之區間方法,以提供在實際應用之參考。本研究並以兩組實際資料-轉基因玉米CBH 351的檢測與甜椒種子病毒PMMoV的檢測為例,來加以闡述所提供統計方法之應用,相信此研究結果在農業實際應用上將有所助益。 |
英文摘要 | In the investigation of agricultural science, provided that the level of infection in the population is not too large, considerable gains in efficiency may be made by group testing. Group testing is a method of pooling a number of units together and performing a single test on the resulting group. The procedure is particularly useful when the number of positive units is expected to be low and obtaining test material is cheap, but testing itself is expensive. In most cases, the classification of groups is binary, if a test is positive; it is assumed that at least one of the plants in the group is positive. Otherwise, it is assumed that all the plants are negative. When threshold of detection exists in analytical facilities and the group size is increased, it might lead to a false negative result, the probability of failing to detect defective items when in reality defective items exist. The objective of this study is to estimate the probability of defective individuals for any group size by using four confidence intervals (Wald, Likelihood Ratio, Score, and Exact methods) when a threshold of detection exists by using group testing. To evaluate and investigate the pros and cons of the four confidence intervals, coverage probability, expected width, and mean absolute deviation were used as the criteria. Finally, optimal confidence intervals will be recommended for providing the practical use. Moreover, the methods of this study were demonstrated and tested on the detections of genetically modified corn CBH 351 and a seed-transmittable "Pepper mild mottle virus" (PMMoV) in pepper seeds. We believe that the results of the study will be useful for the practical application in agricultural science. |
本系統中英文摘要資訊取自各篇刊載內容。