SellTrend: Inter-Attribute Visual Analysis of Temporal Transaction Data

We present a case study of our experience designing SellTrend, a visualization system for analyzing airline travel purchase requests. The relevant transaction data can be characterized as multi-variate temporal and categorical event sequences, and the chief problem addressed is how to help company analysts identify complex combinations of transaction attributes that contribute to failed purchase requests. SellTrend combines a diverse set of techniques ranging from time series visualization to faceted browsing and historical trend analysis in order to help analysts make sense of the data. We believe that the combination of views and interaction capabilities in SellTrend provides an innovative approach to this problem and to other similar types of multivariate, temporally driven transaction data analysis. Initial feedback from company analysts confirms the utility and benefits of the system.     

Parallelisation of storage cell flood models using OpenMP

This paper describes the implementation and benchmarking of a parallel version of the LISFLOOD-FP hydraulic model based on the OpenMP Application Programming Interface. The motivation behind the study was that reducing model run time through parallelisation would increase the utility of such models by expanding the domains over which they can be practically implemented, allowing previously inaccessible scientific questions to be addressed. Parallel speedup was calculated for 13 models distributed over seven study sites and implemented on one, two, four and in selected cases eight processor cores. The models represent a range of previous applications from large area, coarse resolution models of the Amazon, to fine resolution models of urban areas, to orders of magnitude smaller models of rural floodplains. Parallel speedups were greater for larger model domains, especially for models with over 0.2–0.4 million cells where parallel efficiencies of up to 0.75 on four and eight cores were achieved. A key advantage of using OpenMP and an explicit rather than implicit model was the ease of implementation and minimal code changes required to run simulations in parallel.     

Finding the number of clusters in ordered dissimilarities

As humans, we have innate faculties that allow us to efficiently segment groups of objects. Computers, to some degree, can be programmed with similar categorical capabilities, which stem from exploratory data analysis. Out of the various subsets of data reasoning, clustering provides insight into the structure and relationships of input samples situated in a number of distributions. To determine these relationships, many clustering methods rely on one or more human inputs; the most important being the number of distributions, c, to seek. This work investigates a technique for estimating the number of clusters from a general type of data called relational data. Several numerical examples are presented to illustrate the effectiveness of the proposed method.     

A Randomized Algorithm for Online Unit Clustering

In this paper, we consider the online version of the following problem: partition a set of input points into subsets, each enclosable by a unit ball, so as to minimize the number of subsets used. In the one-dimensional case, we show that surprisingly the naïve upper bound of 2 on the competitive ratio can be beaten: we present a new randomized 15/8-competitive online algorithm. We also provide some lower bounds and an extension to higher dimensions.     

Case-Based Learning: Predictive Features in Indexing

Interest in psychological experimentation from the Artificial Intelligence community often takes the form of rigorous post-hoc evaluation of completed computer models. Through an example of our own collaborative research, we advocate a different view of how psychology and AI may be mutually relevant, and propose an integrated approach to the study of learning in humans and machines. We begin with the problem of learning appropriate indices for storing and retrieving information from memory. From a planning task perspective, the most useful indices may be those that predict potential problems and access relevant plans in memory, improving the planner's ability to predict and avoid planning failures. This predictive features hypothesis is then supported as a psychological claim, with results showing that such features offer an advantage in terms of the selectivity of reminding because they more distinctively characterize planning situations where differing plans are appropriate.We present a specific case-based model of plan execution, RUNNER, along with its indices for recognizing when to select particular plans—appropriateness conditions—and how these predictive indices serve to enhance learning. We then discuss how this predictive features claim as implemented in the RUNNER model is then tested in a second set of psychological studies. The results show that learning appropriateness conditions results in greater success in recognizing when a past plan is in fact relevant in current processing, and produces more reliable recall of the related information. This form of collaboration has resulted in a unique integration of computational and empirical efforts to create a model of case-based learning.     

Can twice weekly hemodialysis expand patient access under resource constraints?

The convention of prescribing hemodialysis on a thrice weekly schedule began empirically when it seemed that this frequency was convenient and likely to treat symptoms for a majority of patients. Later, when urea was identified as the main target and marker of clearance, studies supported the prevailing notion that thrice weekly dialysis provided appropriate clearance of urea. Today, national guidelines on hemodialysis from most countries recommend patients receive at least thrice weekly therapy. However, resource constraints in low‐ and middle‐income countries (LMIC) have resulted in a substantial proportion of patients using less frequent hemodialysis in these settings. Observational studies of patients on twice weekly dialysis show that twice weekly therapy has noninferior survival rates compared with thrice weekly therapy. In fact, models of urea clearance also show that twice weekly therapy can meet urea clearance “targets” if patients have significant residual function or if they follow a protein‐restricted diet, as may be common in LMIC. Greater reliance on twice weekly therapy, at least at the start of hemodialysis, therefore has potential to reduce health care costs and increase access to renal replacement therapy in low‐resource settings; however, randomized control trials are needed to better understand long‐term outcomes of twice versus thrice weekly therapy.     

A computational framework for scale‐bridging in multi‐scale simulations

A computational framework for scale‐bridging in multi‐scale simulations is presented. The framework enables seamless combination of at‐scale models into highly dynamic hierarchies to build a multi‐scale model. Its centerpiece is formulated as a standalone module capable of fully asynchronous operation. We assess its feasibility and performance for a two‐scale model applied to two challenging test problems from impact physics. We find that the computational cost associated with using the framework may, as expected, become substantial. However, the framework has the ability of effortlessly combining at‐scale models to render complex multi‐scale models. The main source of the computational inefficiency of the framework is related to poor load balancing of the lower‐scale model evaluation We demonstrate that the load balancing can be efficiently addressed by recourse to conventional load‐balancing strategies. Copyright © 2016 John Wiley & Sons, Ltd.