Discriminant Analysis of Plasma Fusion Data

Recent progress on the development and analysis of the ITPA global H-mode confinement database

“…However, until a more reliable method of assessing the measurement errors is established, it is not clear if a given EIV scaling should be used in preference to an OLS scaling. Besides OLS and EIV, Bayesian methods [90,91] have also been applied to confinement scalings, for a database comprised of data taken from stellarators [92]. Some preliminary Bayesian analysis of the H-mode Database has also been performed [93], and this work is expected to continue in the coming years.…”

A high field side pellet penetration depth scaling derived for ASDEX Upgrade

“…An overview of the parameters that are deemed important for the scaling analysis is presented in table 1. In this table, notation and the units of the parameters used are also indicated, as well as their parameter range and some univariate statistics [12]. The estimated values of the standard deviation, skewness and kurtosis displayed are defined as follows [13], where x = 1/N • i x i denotes the empirical mean of the distribution:…”

“…In table 2 the relationship between the parameters is described in the form of the correlation matrix [12]. It is to be noted that the dataset is not entirely homogeneous and uncorrelated.…”

“…Also the chance of overfitting the dataset is prominent in such cases. In our case three model selection algorithms described in [12] are used for this study to find a good subset of variables. These methods are the following.…”

Reflections on Fourteen Cryptic Issues Concerning the Nature of Statistical Inference*

“…For this topic, the reader is referred to [152,27,72,106,76,77,78,87]. Related to this is the question whether or not a particular type of confinement regime is reached at the reference operating point [72,73,117,77,133], and the prediction of plasma temperature and density profiles [71,94,77,97]. Some statistical areas the authors have been involved in are variable selection and model screening, discriminant analysis, growth curves, multivariate analysis, survival analysis, meta-analysis, design of experiments, size and shape analysis, distributional inference and other foundational issues, such as 'structuring the inferential contest' [41,3].…”