MLwiN 命令介面

MLwiN has two components, a front end and a back end. The front end provides the GUI based interface to the user. The back end, is primarily the original MLn DOS command based program with an interfacing module added, which allows it to receive requests for actions to be executed from the front end and to pass back the results of actions to the front end. There is a command interface window which allows the user to issue commands as in MLn.
　

The front end's requests are mediated by sending commands or groups of commands to the back end. For example, adding explanatory variables by clicking on objects in the equations window will result in the EXPL command being sent to the back end. Conversely if the user issues the EXPL command, in the command interface window or by obeying a macro containing the EXPL command the equations window will be properly updated.
　

The MLwiN help system fully documents these commands together with new commands which allow manipulation of aspects of the GUI, including the design of menu dialogue boxes, and is especially for users interested in writing macros. In release 1.0 it is necessary to use the command interface for the fitting of certain kinds of more advanced multilevel models such as those for time series and survival analysis.
　

MLwiN 2.01 版本提升

有來自1.10 版本的對MLwiN 2.01 版本的幾個提升。

統計過程

1. MCMC 估計已經被提升相當大(從網站看見新手冊可下載) . 下列特徵現下已經被補充說了(注意到︰ 其中一些在1.20 版本裡也是可提供的) :

2. DIC 診斷下是可提供的(Spiegelhalter et al.,2002. 模型複雜性的Bayesian 措施和(用討論)適合 . 王室的統計社會，B的雜誌， ( 64 ) : 583-640).

3. 多層的要素分析模型可能裝有倍數(被有相互關聯或者不相關) 在每個水準的原素。

 
4. Multicategory預訂和無次序的回應模型可以被適合。

 
5. 連續和二進製的回應的多元的混合物可以被適合。

6. 複雜的層1 模型可以被適合。

7. 多元的模型，包括有錯過回應的那些，可以被適合。

8. 在預言者裡的測量誤差的調整可以被適合。

9. 十字架分類模型可以被適合。

10. 多個會員模型可以被適合。

11. 在第1步的自回歸的架構可以被適合。