ISSN:0266-4763    

DOI:10.1080/02664769100000001

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

The shortcomings of conventional statistical packages are discussed to illustrate the need to develop software which is able to exhibit a greater degree of statistical expertise, thereby reducing the misuse of statistical methods by those not well versed in the art of statistical analysis. Up to now the majority of the research into developing knowledge-based statistical software has concentrated on moving away from conventional architectures by adopting what can be termed expert systems approaches. This paper proposes an approach which is based upon the concept of semantic modelling. By representing some of the semantic meaning of data, it is conceived that a system could examine a request to apply a statistical technique and check if the use of the chosen technique was semantically sound, i.e. will the results obtained be meaningful. Current systems, in contrast, can only perform what can be considered as syntactic checks. The prototype system that has been implemented to explore the feasibility of such an approach is presented; the system has been designed as an enhanced variant of a conventional style statistical package.

ISSN:0266-4763    

DOI:10.1080/02664769100000002

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

If several agents must collectively choose one of several possible acts (or compute a single joint control strategy) and if they disagree about the probable

ISSN:0266-4763    

DOI:10.1080/02664769100000003

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

The authors have developed an expert system for the analysis of variance to provide guidance to the data analyst on preparing data, selecting statistical methods and choosing computer programs. This system queries the user on the type of data, sampling method and design considerations. It then suggests an appropriate analysis of variance routine and statistical package. The system provides help in understanding the computer output and with model evaluation. The primary user is a scientist with some knowledge of statistics. This system gives the user easy access to statistical expertise and frees the consultant from some important but routine tasks. It may also be a tool to train users and statistical consultants. A significant benefit of this exercise was an increased understanding of the statistical strategies used by consultants and the complexity of the statistical consulting process between humans. This paper will focus on the problem formulation, acquisition of the knowledge base, the model structure and educational benefits.

ISSN:0266-4763    

DOI:10.1080/02664769100000004

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

A method is given of choosing k-way partitions (where 2 ≤ k ≤ (number of categories of predictor variable)) in classification or decision tree analyses. The method, like that proposed by Kass, chooses the best partition on the basis of statistical significanceand uses the Bonferroni inequality to calculate the significance. Unlike Kass's algorithm, the algorithm does not favour simple partitions (low values of k) nor does it discriminate against free-type (no restriction on order of values) predictor variables with many categories. A method of adjusting the significance for the number of predictor variables and of using multiple comparisons to put an upper bound on the significance is given. Monte Carlo tests show that the algorithm gives slightly conservative tests of significance for both small and large samples and does not favour one type of predictor variable over another. The algorithm is incoroporated in a PC software program, Knowledgeseeker,which is briefly described.

ISSN:0266-4763    

DOI:10.1080/02664769100000005

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

The graphical belief model is a versatile tool for modeling complex systems. The graphical structure and its implicit probabilistic and logical independence conditions define the relationships between many of the variables of the problem. The graphical model is composed of a collection of local models:models of both interactions between the variables sharing a common hyperedge and information about single variables. These local models can be constructed with either probability distributions or belief functions. This paper takes the latter approach and describes simple models for univariate and multivariate belief functions. The examples are taken from both reliability and knowledge representation problems.

ISSN:0266-4763    

DOI:10.1080/02664769100000006

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

The process of data analysis can be divided into stages. The opinions of 20 statisticians on the considerations that are relevant to this decision have been studied. Nineteen concepts were identified which can be classijied into five groups:the analysis question; characteristics of the data collection; characteristics of the data; conditions for and influences on data analysis; conditions for and influences on statistical consultation. Economists mention considerations different to those mentioned by social scientists. Together the considerations form an exhaustive list of concepts that are relevant to the design of computerized support in statistics.

ISSN:0266-4763    

DOI:10.1080/02664769100000007

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

Most approaches to applying knowledge-based techniques for data analyses concentrate on the context-independent statistical support. EXPLORA however is developed for the subject-specific interpretation with regard to the contents of the data to be analyzed (i.e. content interpretation). Therefore its knowledge base includes also the objects and semantic relations of the real system that produces the data. In this paper we describe the functional model representing the process of content interpretation, summarize the software architecture of the system and give some examples of its applications by pilot-users in survey analysis. EXPLORA addresses applications with data produced regularly which have to be analyzed in a routine way. The system systematically searches for statistical results (facts) to detect relations which possibly could be overlooked by a human analyst. On the other hand EXPLORA will help overcome the large bulk of information which today is usually still produced when presenting the data. Therefore a second knowledge process of content interpretation consists in discovering messages about the data by condensing the facts. Approaches for inductive generalization which have been developed for machine learning are utilized to identify common values of attributes of the objects to which the facts relate. At a later stage the system searches for interesting facts by applying redundancy rules and domaindependent selection rules. EXPLORA formulates the messages in terms of the domain, groups and orders them and even provides flexible navigations in the fact spaces.

ISSN:0266-4763    

DOI:10.1080/02664769100000008

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

We studied the way in which experienced statisticians handle the output of a correspondence analysis program, in order to investigate the possibility of computerized support for the interpretation of complex output from statistical software. Four experts participated in this study, all with experience in consultation and with a high level of theoretical knowledge on the subject. In ‘thinking-aloud sessions' they tried to reach conclusions from two forms of output (first without and then with indication of the meaning of variables). We present results on rule-based reasoning by the experts. Suggestions are given for the application of these results in the construction of an intelligent user interface for correspondence analysis.

ISSN:0266-4763    

DOI:10.1080/02664769100000009

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor

摘要:

This paper first describes a program AGREE calculating many variants of coefficients for interobserver agreement. A pilot program MOCK was written, aimed at helping unitiated users of AGREE to select the most appropriate coeflcient, given the data type and the research goal. It is a mock-up version of the data entrance and analysis sections of AGREE, to which are added some menus and a knowledge-based CONSULTANT system questioning the user. Results of a small experiment with four variants of the CONSULTANT are presented. This leads to a discussion of desirable features for this kind of help programs and preprocessors for specialized statistical software.

ISSN:0266-4763    

DOI:10.1080/02664769100000010

出版商:Carfax Publishing Company    

年代:1991

数据来源: Taylor