頁籤選單縮合
題名 | 偽概似函數估計量在二元馬可夫隨機域上的應用=Applications of Pseudo-Likelihood Estimation on Binary Markov Random Fields |
---|---|
作者 | 林培生; 黃怡紋; | 書刊名 | 中國統計學報 |
卷期 | 44:1 民95.03 |
頁次 | 頁91-107 |
分類號 | 319.16 |
關鍵詞 | 馬可夫隨機域; 偽概似函數估計; 自我羅吉斯模型; Markov random field; Pseudo-likelihood estimation; Auto-logistic model; |
語文 | 中文(Chinese) |
中文摘要 | 在本文中,我們利用自我羅吉斯模型和偽概似函數估計法來分析具有相關性的二元資料。雖然最大偽概似函數估計量的近似常態分佈已被證明 (Gidas, 1993),但其近似理論上的變異數與樣本變異數在二元隨機域之下仍存在著相當的差異。在本文中,我們將對Gidas(1993)由Fisher information矩陣而得的近似變異數做簡單修正,並使用模擬實驗來檢驗其在二元馬可夫隨機域的果效。同時,我們將應用此結果於二元自我羅吉斯模型以分析在愛爾蘭中部蝴蝶分佈情形。 |
英文摘要 | In this paper, we use an auto-logistic model and pseudo-likelihood estimation to analyze binary correlated data. Although the asymptotic normality of pseudo-likelihood estimates has been proved (Gidas, 1993), the difference between the proposed asymptotic variance and the sample variance is still huge in binary Markov random fields. In this paper, we first use the Taylor expansion to derive a simple formula to modify the asymptotic variance, and then study its performance by simulations. A data set from the ringlet butterfly in the center of Ireland is used to illustrate the method. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。