Parameter selection in modified histogram estimates

A multivariate modified histogram density estimate depending on a reference density g and a partition P has been proved to have good consistency properties according to several information theoretic criteria. Given an i.i.d. sample, we show how to select automatically both g and P so that the expected L ₁ error of the corresponding selected estimate is within a given constant multiple of the best possible error plus an additive term which tends to zero under mild assumptions. Our method is inspired by the combinatorial tools developed by Devroye and Lugosi [Devroye, L. and Lugosi, G., 2001, Combinatorial Methods in Density Estimation (New York, NY: Springer–Verlag)] and it includes a wide range of reference density and partition models. Results of simulations are also presented.     

On Combining Wavelets Expansion and Sparse Linear Models for Regression on Metabolomic Data and Biomarker Selection

Wavelet thresholding of spectra has to be handled with care when the spectra are the predictors of a regression problem. Indeed, a blind thresholding of the signal followed by a regression method often leads to deteriorated predictions. The scope of this article is to show that sparse regression methods, applied in the wavelet domain, perform an automatic thresholding: the most relevant wavelet coefficients are selected to optimize the prediction of a given target of interest. This approach can be seen as a joint thresholding designed for a predictive purpose. The method is illustrated on a real world problem where metabolomic data are linked to poison ingestion. This example proves the usefulness of wavelet expansion and the good behavior of sparse and regularized methods. A comparison study is performed between the two-steps approach (wavelet thresholding and regression) and the one-step approach (selection of wavelet coefficients with a sparse regression). The comparison includes two types of wavelet bases, various thresholding methods, and various regression methods and is evaluated by calculating prediction performances. Information about the location of the most important features on the spectra was also obtained and used to identify the most relevant metabolites involved in the mice poisoning.     

Inequality decomposition values: the trade-off between marginality and efficiency

This paper presents a general procedure for decomposing income inequality measures by income sources. The methods of decomposition proposed are based on the Shapley value and extensions of the Shapley value of transferable utility cooperative games. In particular, we find that Owen’s value can find an interesting application in this context.We show that the axiomatization by the potential of Hart and Mas-Colell remains valid in the presence of the domain restriction of inequality indices. We also examine the properties of these decomposition rules and perform a comparison with Shorrocks’ decomposition rule properties.     

Painting the Prison 'Red': Constructing and Experiencing Aboriginal Identities in Prison

Dominant Western paradigms of the social work profession havelargely failed to integrate Aboriginal traditional knowledgesand practices on healing and helping. This paper contributesto the promotion of a context-based approach to social workin prison by examining Aboriginality from both institutionaland individual points of view. Drawing on documentary analysesand interviews with Aboriginal women prisoners in Canada, thepaper sheds light on the prison’s endorsement of a hegemonicvision of Aboriginality, and on social work practitioners’inclination to adhere to it. Conversely, we argue that Aboriginalwomen prisoners negotiate their passage into prison throughAboriginal self-identification configurations that often havelittle in common with the prison’s vision of Aboriginality.Service delivery in prison may be enhanced by considering individualmodes of resisting identity-based oppression in prison, andby challenging prisons’ master narrative on Aboriginality.     

Sharing beliefs and the absence of betting in the Choquet expected utility model

Choquet expected utility maximizers tend to behave in a more “cautious” way than Bayesian agents, i. e. expected utility maximizers. We illustrate this phenomenon in the particular case of betting behavior. Specifically, consider agents who are Choquet expected utility maximizers. Then, if the economy is large, Pareto optimal allocations provide full insurance if and only if the agents share at least on prior, i. e., if the intersection of the core of the capacities representing their beliefs is non empty. In the expected utility case, this is true only if they have a common prior. Received: July 2000; revised version: May 2001     

Social Protection: Values to be Defended!

Social protection is a part of human rights. I therefore believe it is important to react to the many reproaches levelled at it, which are undermining its legitimacy. This involves defending social protection, first and foremost in the name of its mission, its fundamental objectives and, consequently, the values upon which it is founded. After briefly reviewing the content of these values, I emphasize the fact that they should be promoted and cultivated at least as much as the economic values that are dominant in the world today.     

Small-sample results for maximum-likelihood estimation in branching processes

In this article, small sample properties of the maximum-likelihood estimator (m.l.e.) for the offspring distribution (pk) and its mean m are considered in the context of the simple branching process. A representation theorem is given for the m.l.e. of (Pk) from which the m.l.e. of m is obtained. The case where p0 + p1 + p2 = 1 is studied in detail: numerical results are given for the exact bias of these estimators as a function of the age of the process; a curve fitting analysis expresses the bias of m? as a function of the mean and the variance of the offspring distribution and finally an “approximate m.l.e.” for (pk) is given.