廣義成長曲線模型的預測分析

Abstract

由於曲線模型於廣義成長的預測分佈的複雜性, 我們提出幾種可行的近似 預測分佈。藉由Gibbs sampling所模擬的預測分佈, 和近似預測分佈比 較, 我們可以得到最佳的近似分佈。 Due to the complexity of predictive distributions forthe generalized growth curve model, we propose some useful approximations.Using the Gibbs sampling, we can compare the simulations with theapproximate predictive distributions to obtain the best approximationamong the candidates considered.