From d20e2e505d7d615b1d6da06cbc3b29c2472354b0 Mon Sep 17 00:00:00 2001 From: AJ Rossini Date: Wed, 24 Mar 2010 14:03:56 +0100 Subject: [PATCH] update to rationale for this new system. Signed-off-by: AJ Rossini --- Doc/papers/CLS-philosophy.tex | 32 ++++++++++++++++++++++++++++---- 1 file changed, 28 insertions(+), 4 deletions(-) diff --git a/Doc/papers/CLS-philosophy.tex b/Doc/papers/CLS-philosophy.tex index 4a93778..c14e4f1 100644 --- a/Doc/papers/CLS-philosophy.tex +++ b/Doc/papers/CLS-philosophy.tex @@ -27,16 +27,27 @@ approach dating back to the dawn of the computing era. \section{Components of a procedure} \label{sec:components} +Statistics consists of a range of procedures that can be applied to +make decisions. However, the range of procedures and resulting +interpretation makes it difficult to actually drive a cook-book style +approach. But it remains that there is a color-palette of procedures +which can be used to support decision making from collected +datatsets. + define a statistical procedure as a decision-making approach which entails the intertwining of formal and informal structure. -Components: +Components, first pass \begin{enumerate} \item \label{statproc-decision} Decision to make -\item \label{statproc-assessment} Assessment approach to use - (some are inherently different, others just look different) +\item \label{statproc-assessment} Assessment approach to use (some are + inherently different, others just look different): example, T-test + is a simplified ANOVA, so look different. Or ML-based (with a + hill-climb) vs. ML with a LS fit (look different) vs. bayesian + Linear regression (inherently different). \item \label{statproc-normalization} Normalization of the problem for - assessment/comparison with other reference behaviours + assessment/comparison with other reference behaviours or + probabilistic processes. \item \label{conclusion} Type of conclusion desired, and instance of that conclusion (when data is present) \end{enumerate} @@ -45,6 +56,15 @@ This forms an \textit{abstract class} of a procedure, which can be represented by a real class, which can then be instantiated through the application of data. +Components from Gelman: +\begin{enumerate} +\item design or analysis state (0th order) +\item measurement +\item comparison +\item ??normalization?? +\end{enumerate} +to characterize the application of a statistical procedure. + \subsection{Decision} \label{sec:components:decision} @@ -59,6 +79,10 @@ Questions could be: \item what is the strength of the difference? \end{itemize} +One component (from Gelman's dicotomy) -- comparison. And decision +could equal ( comparsion, extremeness from expectation ). + + \subsection{Core Assessment} \label{sec:components:assessment} -- 2.11.4.GIT