Visualization of diversity in large multivariate data sets

Understanding the diversity of a set of multivariate objects is an important problem in many domains, including ecology, college admissions, investing, machine learning, and others. However, to date, very little work has been done to help users achieve this kind of understanding. Visual representation is especially appealing for this task because it offers the potential to allow users to efficiently observe the objects of interest in a direct and holistic way. Thus, in this paper, we attempt to formalize the problem of visualizing the diversity of a large (more than 1000 objects), multivariate (more than 5 attributes) data set as one worth deeper investigation by the information visualization community. In doing so, we contribute a precise definition of diversity, a set of requirements for diversity visualizations based on this definition, and a formal user study design intended to evaluate the capacity of a visual representation for communicating diversity information. Our primary contribution, however, is a visual representation, called the Diversity Map, for visualizing diversity. An evaluation of the Diversity Map using our study design shows that users can judge elements of diversity consistently and as or more accurately than when using the only other representation specifically designed to visualize diversity.     

Fault diagnosis of car assembly line based on fuzzy wavelet kernel support vector classifier machine and modified genetic algorithm

This paper presents a new version of fuzzy wavelet support vector classifier machine to diagnosing the nonlinear fuzzy fault system with multi-dimensional input variables. Since there exist problems of finite samples and uncertain data in complex fuzzy fault system, the input and output variables are described as fuzzy numbers. Then by integrating the fuzzy theory, wavelet analysis theory and v-support vector classifier machine, fuzzy wavelet v-support vector classifier machine (FWv-SVCM) is proposed. To seek the optimal parameters of FWv-SVCM, genetic algorithm (GA) is also applied to optimize unknown parameters of FWv-SVCM. A diagnosing method based on FWv-SVCM and GA is put forward. The results of the application in car assembly line diagnosis confirm the feasibility and the validity of the diagnosing method. Compared with the traditional model and other SVCM methods, FWv-SVCM method requires fewer samples and has better diagnosing precision.     

Online bin stretching with three bins

Online bin stretching is a semi-online variant of bin packing in which the algorithm has to use the same number of bins as an optimal packing, but is allowed to slightly overpack the bins. The goal is to minimize the amount of overpacking, i.e., the maximum size packed into any bin. We give an algorithm for online bin stretching with a stretching factor of \(11/8 = 1.375\) for three bins. Additionally, we present a lower bound of \(45/33 = 1.\overline{36}\) for online bin stretching on three bins and a lower bound of 19/14 for four and five bins that were discovered using a computer search.     

Abstract processes of place/transition systems

A well-known problem in Petri net theory is to formalise an appropriate causality-based concept of process or run for place/transition systems. The so-called individual token interpretation, where tokens are distinguished according to their causal history, giving rise to the processes of Goltz and Reisig, is often considered too detailed. The problem of defining a fully satisfying more abstract concept of process for general place/transition systems has so-far not been solved. In this paper, we recall the proposal of defining an abstract notion of process, here called BD-process, in terms of equivalence classes of Goltz-Reisig processes, using an equivalence proposed by Best and Devillers. It yields a fully satisfying solution for at least all one-safe nets. However, for certain nets which intuitively have different conflicting behaviours, it yields only one maximal abstract process. Here we identify a class of place/transition systems, called structural conflict nets, where conflict and concurrency due to token multiplicity are clearly separated. We show that, in the case of structural conflict nets, the equivalence proposed by Best and Devillers yields a unique maximal abstract process only for conflict-free nets. Thereby BD-processes constitute a simple and fully satisfying solution in the class of structural conflict nets.     

Optimal combination forecasts for hierarchical time series

In many applications, there are multiple time series that are hierarchically organized and can be aggregated at several different levels in groups based on products, geography or some other features. We call these “hierarchical time series”. They are commonly forecast using either a “bottom-up” or a “top-down” method.In this paper we propose a new approach to hierarchical forecasting which provides optimal forecasts that are better than forecasts produced by either a top-down or a bottom-up approach. Our method is based on independently forecasting all series at all levels of the hierarchy and then using a regression model to optimally combine and reconcile these forecasts. The resulting revised forecasts add up appropriately across the hierarchy, are unbiased and have minimum variance amongst all combination forecasts under some simple assumptions.We show in a simulation study that our method performs well compared to the top-down approach and the bottom-up method. We demonstrate our proposed method by forecasting Australian tourism demand where the data are disaggregated by purpose of travel and geographical region.     

Product Focused Software Process Improvement: Concepts and Experiences from Industry

Management problems in the development of software have been addressed over the last years by a strong focus on the improvement of the development processes. Software process improvement (SPI) activities are characterized by an internal focus on a software development department and its procedures. However, the quality of the product is hardly addressed in software process improvement programs. This paper presents the application of a model for product focused SPI (P-SPI) and describes experiences with this model in practice. The main conclusions are that P-SPI puts products to be developed in a central position in improvement programs, results in fulfilling specific quality goals of a company and project, and industrial experiments show interesting benefits. Applying the approach in industrial projects showed that P-SPI is relatively cheap and gives fast results and high benefits.     

Quality through Managed Improvement and Measurement (QMIM): Towards a Phased Development and Implementation of a Quality Management System for a Software Company

The paper describes results of a longitudinal study of developments in the area of software product and process quality improvement within a Hungarian software company, IQSOFT Ltd. This company has been active in this area since 1993, trying to build, introduce and maintain an efficiently working quality management system which, e.g., fulfils the ISO 9001 requirements, allows steady software process improvement and, at the same time, conforms to company's own needs. Over the last eight years five phases could be distinguished. Each phase is described shortly, following the same structure, namely: basic starting points, key problem areas, literature consulted, activities and design executed, reflections on what happened and why. The lessons resulting from the analysis of this case have been formulated in terms of guidelines. We feel that these are applicable to any low maturity software development organisation embarking on a product or process quality improvement endeavour. These guidelines are developed around a framework containing the basic issues of software production (project management, technical processes and products). The guidelines advocate a careful step-by-step development of definitions, quality characteristics, and metrics related to these objects while at the same time developing and introducing the associated process.     

The use of NDVI and its Derivatives for Monitoring Lake Victoria’s Water Level and Drought Conditions

Normalized Difference Vegetation Index (NDVI), which is a measure of vegetation vigour, and lake water levels respond variably to precipitation and its deficiency. For a given lake catchment, NDVI may have the ability to depict localized natural variability in water levels in response to weather patterns. This information may be used to decipher natural from unnatural variations of a given lake’s surface. This study evaluates the potential of using NDVI and its associated derivatives (VCI (vegetation condition index), SVI (standardised vegetation index), AINDVI (annually integrated NDVI), green vegetation function (F _g ), and NDVIA (NDVI anomaly)) to depict Lake Victoria’s water levels. Thirty years of monthly mean water levels and a portion of the Global Inventory Modelling and Mapping Studies (GIMMS) AVHRR (Advanced Very High Resolution Radiometer) NDVI datasets were used. Their aggregate data structures and temporal co-variabilities were analysed using GIS/spatial analysis tools. Locally, NDVI was found to be more sensitive to drought (i.e., responded more strongly to reduced precipitation) than to water levels. It showed a good ability to depict water levels one-month in advance, especially in moderate to low precipitation years. SVI and SWL (standardized water levels) used in association with AINDVI and AMWLA (annual mean water levels anomaly) readily identified high precipitation years, which are also when NDVI has a low ability to depict water levels. NDVI also appears to be able to highlight unnatural variations in water levels. We propose an iterative approach for the better use of NDVI, which may be useful in developing an early warning mechanisms for the management of lake Victoria and other Lakes with similar characteristics.     

A sparse Gaussian process regression model for tourism demand forecasting in Hong Kong

In recent years, Gaussian process (GP) models have been popularly studied to solve hard machine learning problems. The models are important due to their flexible non-parametric modeling abilities using Mercer kernels and the Bayesian framework for probabilistic inference. In this paper, we propose a sparse GP regression (GPR) model for tourism demand forecasting in Hong Kong. The sparsification procedure of the GPR model not only decreases the computational complexity but also improves the generalization ability. We experiment the proposed model with monthly demand data that are relevant to Hong Kong’s tourism industry, and compare the performance of the sparse GPR model with those of various kernel-based models to show its effectiveness. The proposed sparse GPR model shows that its forecasting capability outperforms those of the ARMA model and the two state-of-the-art SVM models.