An incremental least squares algorithm for large scale linear classification

In this work we consider the problem of training a linear classifier by assuming that the number of data is huge (in particular, data may be larger than the memory capacity). We propose to adopt a linear least-squares formulation of the problem and an incremental recursive algorithm which requires to store a square matrix (whose dimension is equal to the number of features of the data). The algorithm (very simple to implement) converges to the solution using each training data once, so that it effectively handles possible memory issues and is a viable method for linear large scale classification and for real time applications, provided that the number of features of the data is not too large (say of the order of thousands). The extensive computational experiments show that the proposed algorithm is at least competitive with the state-of-the-art algorithms for large scale linear classification.     

An approximation algorithm for the Euclidean incremental median problem

In the incremental version of the $k$-median problem, we find a sequence of facility sets $F_1\subseteq F_2\subseteq \cdots \subseteq F_n$, where each $F_k$ contains at most $k$ facilities. This sequence is said to be $\delta$-competitive if the cost of each $F_k$ is at most $\delta$ times the optimum cost of $k$ facilities. The best deterministic (randomized) algorithm available for the metric space has a competitive ratio of $8$ ($7.656$). The best one for the one-dimensional problem finds a $5.828$-competitive sequence. We give a $7.076$-competitive solution for the high-dimensional Euclidean space.     

A framework for incremental generation of closed itemsets

Association rule mining from a transaction database (TDB) requires the detection of frequently occurring patterns, called frequent itemsets (FIs), whereby the number of FIs may be potentially huge. Recent approaches for FI mining use the closed itemset paradigm to limit the mining effort to a subset of the entire FI family, the frequent closed itemsets (FCIs). We show here how FCIs can be mined incrementally yet efficiently whenever a new transaction is added to a database whose mining results are available. Our approach for mining FIs in dynamic databases relies on recent results about lattice incremental restructuring and lattice construction. The fundamentals of the incremental FCI mining task are discussed and its reduction to the problem of lattice update, via the CI family, is made explicit. The related structural results underlie two algorithms for updating the set of FCIs of a given TDB upon the insertion of a new transaction. A straightforward method searches for necessary completions throughout the entire CI family, whereas a second method exploits lattice properties to limit the search to CIs which share at least one item with the new transaction. Efficient implementations of the parsimonious method is discussed in the paper together with a set of results from a preliminary study of the method's practical performances.     

Estimating variance components in linear models

Estimation of variance components in linear model theory is presented as an application of estimation of the mean by introducing a dispersion-mean correspondence. Without any further computations, this yields most general representations of minimum variance-minimum bias-invariant quadratic estimates, estimates from MINQUE theory, and Ridge-type estimates of the variance components.     

Bayes estimation in linear models: A coordinate-free approach

The unified theory of Bayes estimation in linear models is presented, using a coordinate-free approach. The results are applied to the problem of linear and quadratic estimation in linear regression model.     

A rough set-based incremental approach for learning knowledge in dynamic incomplete information systems

With the rapid growth of data sets nowadays, the object sets in an information system may evolve in time when new information arrives. In order to deal with the missing data and incomplete information in real decision problems, this paper presents a matrix based incremental approach in dynamic incomplete information systems. Three matrices (support matrix, accuracy matrix and coverage matrix) under four different extended relations (tolerance relation, similarity relation, limited tolerance relation and characteristic relation), are introduced to incomplete information systems for inducing knowledge dynamically. An illustration shows the procedure of the proposed method for knowledge updating. Extensive experimental evaluations on nine UCI datasets and a big dataset with millions of records validate the feasibility of our proposed approach.     

Qualities of goods in the models of production

In this paper the models of production in which the qualities of goods are considered are presented. Such models are completely described by the families of matrices or by a function which has some properties which are natural from the economic point of view. The special cases of the models are the well known linear model of production and the van Neumann-Gale's model.     

Clustering electricity consumers using high‐dimensional regression mixture models

A massive amount of data about individual electrical consumptions are now provided with new metering technologies and smart grids. These new data are especially useful for load profiling and load modeling at different scales of the electrical network. A new methodology based on mixture of high‐dimensional regression models is used to perform clustering of individual customers. It leads to uncovering clusters corresponding to different regression models. Temporal information is incorporated in order to prepare the next step, the fit of a forecasting model in each cluster. Only the electrical signal is involved, slicing the electrical signal into consecutive curves to consider it as a discrete time series of curves. Interpretation of the models is given on a real smart meter dataset of Irish customers.     

A comparison of linear reformulations for multinomial logit choice probabilities in facility location models

In the last decade several papers appeared on facility location problems that incorporate customer demand by the multinomial logit model. Three linear reformulations of the original non-linear model have been proposed so far. In this paper, we discuss these models in terms of solvability. We present empirical findings based on synthetic data.     

A comprehensive analytical approach for policy analysis of system dynamics models

Formal tools to link system dynamics model’s structure to the system modes of behavior have recently become available. In this paper, we aim to expand the use of these tools to perform the model’s policy analysis in a more structured and formal way than the exhaustive exploratory approaches used to date. We consider how a policy intervention (a parameter change) affects a particular behavior mode by affecting the gains of particular feedback loops as well as how it affects the presence of that mode in the variable of interest. The paper demonstrates the utility of considering both of these aspects since the analysis provides an assessment of the overall impact of a policy on a variable and explains why the impact occurs in terms of structural changes in the model. Particularly in the context of larger models, this method enables a much more efficient search for leverage policies, by ranking the influence of each model parameter without the need for multiple simulation experiments.     

A modified Crank-Nicolson scheme with incremental unknowns for convection dominated diffusion equations

A modified Crank-Nicolson scheme based on one-sided difference approximations is proposed for solving time-dependent convection dominated diffusion equations in two-dimensional space. The modified scheme is consistent and unconditionally stable. A priori L² error estimate for the fully discrete modified scheme is derived. With the use of the incremental unknowns preconditioner at each time step, a comparison among several classical numerical schemes has been made and numerical results confirm stability and efficiency of the modified Crank-Nicolson scheme.     

Robustness of F-tests in singular linear models

F-test is the most popular test in the general linear model. However, there is few discussions on the robustness of F-test under the singular linear model. In this paper, the necessary and sufficient conditions of robust F-test statistic are given under the general linear models or their partition models, which allows that the design matrix has deficient rank and the covariance matrix of error is a nonnegative definite matrix with parameters. The main results obtained in this paper include the existing findings of the general linear model under the definite covariance matrix. The usage of the theorems is illustrated by an example.     

Constitution of random intercept and slope model (RISM) as a special case of linear mixed models (LMMs) for repeated measurements data

In this paper, for the aim of modeling variance-covariance structure matrix of the response variables vector in random intercept and slope model (RISM) from linear mixed models (LMMs) for repeated measurements data, 13 different homogeneous and heterogeneous variance-covariance structure models are investigated comparatively in an application from a clinical trial.     

过拟合最小二乘估计的相对效率

讨论线性回归模型中选入过多的解释变量对回归系数估计的影响,给出由过拟合模型求得的最小二乘估计效率的下界。     

On statistical models for regression diagnostics

In regression diagnostics, the case deletion model (CDM) and the mean shift outlier model (MSOM) are commonly used in practice. In this paper we show that the estimates of CDM and MSOM are equal in a wide class of statistical models, which include LSE, MLE, Bayesian estimate andM-estimate in linear and nonlinear regression models; MLE in generalized linear models and exponential family nonlinear models; MLEs of transformation parameters of explanatory variables in a Box-Cox regression models and so on. Furthermore, we study some models, in which, the estimates are not exactly equal but are approximately equal for CDM and MSOM.     

广义线性模型(十)

本讲座是广义线性模型这个题目的一个比较系统的介绍。主要分3部分:建模、统计分析与模型选择和诊断。写作时依据的主要参考资料是L.Fahrmeir等人的《MultivariateStatisticalModelingBasedonGeneralizedLinearModels》。     

广义线性模型(八)

本讲座是广义线性模型这个题目的一个比较系统的介绍。主要分3部分:建模、统计分析与模型选择和诊断。写作时依据的主要参考资料是L.Fahrmeir等人的《MultivariateStatisticalModel ingBasedonGeneralizedLinearModels》     

广义线性模型(一)

本讲座是广义线性模型这个题目的一个比较系统的介绍。主要分 3部分 :建模、统计分析与模型选择和诊断。写作时依据的主要参考资料是L .Fahrmeir等人的《MultivariateStatisticalModelingBasedonGeneralizedLinearModels》。     

广义线性模型(三)

本讲座是广义线性模型这个题目的一个比较系统的介绍。主要分 3部分 :建模、统计分析与模型选择和诊断。写作时依据的主要参考资料是L .Fahrmeir等人的《MultivariateStatisticalModelingBasedonGeneralizedLinearModels》。     

广义线性模型(二)

本讲座是广义线性模型这个题目的一个比较系统的介绍。主要分 3部分 :建模、统计分析与模型选择和诊断。写作时依据的主要参考资料是L .Fahrmeir等人的《MultivariateStatisticalModelingBasedonGeneralizedLinearModels》。