Panagiotis Chountas

An analysis of missing data treatment methods and their application to health care dataset

It is well accepted that many real-life datasets are full of missing data. In this paper we introduce, analyze and compare several well known treatment methods for missing data handling and propose new methods based on Naive Bayesian classifier to estimate and replace missing data. We conduct extensive experiments on datasets from UCI to compare these methods. Finally we apply these models to a geriatric hospital dataset in order to assess their effectiveness on a real-life dataset.

On imprecision intuitionistic fuzzy sets & OLAP - the case for KNOLAP

Traditional data repositories are typically focused on the storage and querying of crisp-precise domains of data. As a result, current commercial data repositories have no facilities for either storing or querying imprecise-approximate data. However, when considering scientific data (i.e. medical data, sensor data etc) value uncertainty is inherited to scientific measurements. In this paper we revise the context of “value uncertainty”, and examine common models related to value uncertainty as part of the OLAP model. We present our approach for extending the OLAP model to include treatment of value uncertainty as part of a multidimensional model inhabited by flexible date and non-rigid hierarchical structures of organisation.

A data warehouse environment for storing and analyzing simulation output data

Discrete event simulation modelling has been extensively used in modelling complex systems. Although it offers great conceptual-modelling flexibility, it is both computationally expensive and data intensive. There are several examples of simulation models that generate millions of observations to achieve satisfactory point and confidence interval estimations for the model variables. In these cases, it is exceptionally cumbersome to conduct the required output and sensitivity analysis in a spreadsheet or statistical package. In this paper, we highlight the advantages of employing data warehousing techniques for storing and analyzing simulation output data. The proposed data warehouse environment is capable of providing the means for automating the necessary algorithms and procedures for estimating different parameters of the simulation. These include initial transient in steady-state simulations and point and confidence interval estimations. Previously developed models for evaluating patient flow through hospital departments are used to demonstrate the problem and the proposed solutions.

A Decision Support System for Measuring and Modelling the Multi-Phase Nature of Patient Flow in Hospitals

Multi-phase models of patient flow offer a practical but scientifically robust approach to the studying and understanding of the different streams of patients cared for by health care units. In this chapter, we put forward a decision support system that is specifically designed to identify the different streams of patient flow and to investigate the effects of the interaction between them by using readily available administrative data. The richness of the data dictate the use of data warehousing and On-Line Analytical Processing (OLAP) for data analysis and pre-processing; the complex and stochastic nature of health care systems suggested the use of discrete event simulation as the decision model. We demonstrate the application of the decision support system by reporting on a case study based on data of patients over 65 with a stroke related illness discharged by English hospitals over a year.

Development of a clinical data warehouse

There is increasing worldwide awareness that bionics and artificial intelligence will play an important role in microbial analysis. An intelligent data-warehouse system consisting of an odour generation mechanism, rapid volatile delivery and recovery system, and a classifier system based on neural networks and genetic algorithms have been applied as part of a microbial analysis. The microbiological warehouse environment has, also adopted the concept of fusion of multiple classifiers dedicated to specific feature parameters. The experimental results confirm the soundness of the presented methods.

Representation of definite, indefinite and infinite temporal information

An information system is used for representing and managing indicative information from multiple sources describing the state of an enterprise. Most information systems model enterprises that are crisp. A crisp enterprise is one that is highly quantifiable; relationships are fixed and attributes are atomic valued. The premises for this paper are precise enterprises, where data are imperfect and multiple conflicting sources of information do exist. In such cases, information can be imperfect and/or temporal, or any possible combination of each two of them. In domains where information is perfect, all information sources are absolutely reliable. In more speculative domains, like diagnosis, information may be asserted relatively to some time intervals in which it is possibly defined and probably believed. In such domains different sources of information may be assigned different degrees of reliability. Value imperfection and/or temporal indeterminacy may cause uncertainty. Temporal information may be recurring, periodical or definite.

Modelling and Representation of Uncertain Temporal Information

The main task of an information system is the representation and management of large amounts of indicative information from multiple sources describing the state of some enterprise. Most conceptual and database models represent enterprises with no imprecise data. Very few approaches in the literature are dealing with imprecise data. In temporal data models and databases, approaches are dealing mainly with precise absolute times. Little consideration has been given to imprecise absolute times or infinite absolute times. No consideration has been given to imprecise infinite absolute times either. Many algebraic models are dealing with temporal or value imperfection, with no description of semantics of uncertain information. There is a need for conceptual models that capture the essential semantics of data imperfection, belief, and the temporal nature of imperfect information. This paper is proposing a conceptual framework that describes the semantics of temporal and incomplete information, and shows how this can be translated to database representation.

Intuitionistic fuzzy XML query matching

As the popularity of XML as a de facto standard for data representation and communication is rising, a need has been identified for efficient XML querying techniques that can overcome the dilemma of diversity in the structure of XML data sources. In this work, we propose our approach of using Intuitionistic Fuzzy Trees (IFTr) to achieve approximate XML query matching, making the query result include not just the XML data trees that exactly match the query, but also the ones that partially match it. Our approach has a potential to return useful query answers while pertaining good performance. Users will have the option to choose between quick and less accurate results or time costing and more accurate results.

Enhancing DWH models with the utilisation of multiple hierarchical schemata

Data Warehouse (DWH) Models are based on static dimensions having single hierarchies. With the evolution of World Wide Web, including external knowledge can enrich those models and provide better results of data analysis. Therefore, it would be useful to have intelligent transformation utilities that can mine various data structures and extract useful knowledge participating in the construction of flexible DWH models. This paper proposes a new approach towards intelligent transformation utilities that will allow the utilisation of multiple hierarchical schemata when defining the dimensions of data-warehouses. By allowing a particular dimension to have multiple but semantically close definitions, we allow same users to query same data with the aid of different semantics. To put it differently users are allowed to change or refine the axis of analysis with respect to a particular query, in their effort to achieve a more meaningful answer. We make use of Intuitionistic Fuzzy Logic to soften the rules of calculating the similarity between different hierarchies which in turn is used to decide if the hierarchies can be included in the definition of data warehouse dimension. Data transformations are used to transform the data from one hierarchy to another. With the aid of external data, some sort of estimation is used to estimate values of new hierarchy levels and then based on the users request the desired hierarchy is used to view an OLAP cube.

An intuitionistic fuzzy component based approach for identifying Web usage patterns

This work presents a framework for Web mining, which is developed to support and assist existing data mining algorithms in order to preliminarily refine browsing pattern with relevant constraints. Intuitionistic fuzzy sets are used to represent the possibility that backward steps are used while searching for the pages of interest.