Wyniki wyszukiwania - Biblioteka Nauki

1

The Board of Directors, Leading Firms and Sectors. An Attempt of Data Mining Based Description

100%

Karbownik K. , Boguta M. , Wrzosek M.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

41-62

EN

The relations between attributes describing the Board of Directors and the economic performance of a relevant company, and that of the sectors in national economy are investigated. The Principal - Agent Theory and the Game Theory cannot be used in satisfactory way to verify economic behavior arising in particular situations, because they operate on general concepts. The paper presents an attempt to reformulate this issue, taking it into account when explaining the process of operational decision making. The usual focus on what the Boards of Directors do, how their composition may affect it, and what should be done to improve their effectiveness is extended to include the issue of criteria used in Board selection process. Search for empirical regularity within director's selection process leads to learn how directors are chosen and to understanding corporate organization and governance.

2

My AI or Five Theses about Artificial Intelligence after its 50th Birthday

100%

Matwin S.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

9-18

EN

AIn a search for lessons learned from 50 years of history of AI, this paper presents a brief, subjective and personal history of the field. It then introduces five theses-prescriptions for what makes good AI research. The theses stem form the author's understanding of successes and failures of the field, and from his own experience as a long-standing and active member of the AI community. The five theses promote practicality, embeddedness, empirical verification, mathematical foundation, and scrutiny.

3

DEVELOPMENT OF DATA MINING MODELS FOR ANALYSIS OF CHARACTERISTICS OF UNEMPLOYED PERSONS AND PREDICTING OF UNEMPLOYMENT DURATION (Polish title below)

100%

Lasek M. , Peczkowski M.

Optimum. Studia Ekonomiczne

|

2012

|

nr 1(55)

119-142

EN

(Polish title: Budowa modeli eksploracji danych (data mining) na potrzeby analiz cech charakteryzujacych osoby bezrobotne i przewidywania pozostania osoba bezrobotna). The article discusses data mining models that can be useful in the analysis of attributes influencing individual unemployment and can help to estimate the probability with which persons with certain characteristics (age, education, gender, family background, family status, family size etc.) are more likely to be unemployed or to lose their job. Three models were constructed, and verified, to accomplish this task: logistic regression, multilayered neural network and decision tree. The study was based on Polish official statistical data. SAS Enterprise Miner software from SAS Institute Inc. was applied for model construction and data analysis.

4

PREDICTION MODELS ENSEMBLE WITH MSE DECOMPOSITION (Agregacja modeli predykcyjnych z wykorzystaniem dekompozycji bledu sredniokwadratowego)

100%

Szupiluk R.

Roczniki Kolegium Analiz Ekonomicznych

|

2009

|

tom 19

163-170

EN

In this paper the author presents the theoretical background for ensemble methods via multivariate decompositions. Having a set of models he treats their results as a multivariate variable with destructive and constructive latent components. The data decorrelation and MSE error decompositions provide to proper destructive components identification and elimination. The elimination of destructive components should improve final prediction. The validity of the approach is verified by energy load prediction problem.

5

On Patterns in Economic Data and Monetary Councils Decisions

100%

Koloch G. , Kozlowski W. , Piekos P.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

187-202

EN

In the paper an attempt to identify decision rules which emulate policy decisions of monetary councils (councils) in Poland, United States and Japan is presented. Policy decisions are defined as decisions concerning target interest rate changes. Generated decision rules emulate therefore monetary authorities' reaction function or policy rules. Accuracy of these rules is measured by the ratio of instances (decisions) correctly classified to the number of all instances under consideration. Generated rules are meaningful. Most of them can be interpreted in terms of stabilizing policy, which stimulates output or decelerates inflation dependent on the evolution of economic situation in the domestic economy. This is consistent with the goal of monetary policy, which is to stabilize prices and (in case of USA) output fluctuations.

6

Influence of the Education on the Level of Unemployment in European Countries

100%

Wrzosek M. , Karbownik K. , Boguta M.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

101-118

EN

The results of an empirical survey on the relationship between unemployment rate and selected macroeconomic indicators are presented. The goal of the analysis was to verify the hypothesis on significant correlation of the level of unemployment and the society's level of education. In addition, an investigation on the strength of the relationship between unemployment level and education linked with labor policy was conducted. Two hypotheses were investigated. The first one states that within the group of 29 surveyed countries separate subgroups, internally coherent can be distinguished. The second hypothesis postulates that the unemployment rate depends stronger on the level of expenditures and other indicators linked to education or development than on active labor market policies.

7

Welfare Determinants: Educational, Occupational and Income Perspective

100%

Wojewnik P. , Zawisza M.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

63-84

EN

In this paper an attempt to verify, whether the empirical data confirm the welfare determinants suggested in economic literature, like e.g. educational, geographical, behavioural features of the individual and his/her parents, see e.g. Wedgwood (1928). The analysis is conducted using machine learning tools and then the obtained results are compared with other studies. To ensure the possibility of such assessment the capability of interpretation in terms of natural language is needed and thus the knowledge in the form of classification rules is presented.

8

Recognition of Households Consumption Preferences' Profiles

100%

Kuszewski T. , Sokolinska A. , Zakrzewska K.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

19-40

EN

This article shows the results of the households' income and expenditures analysis. Special attention was given to the relation between food and alcohol-free beverages in general expenditures and social-demographic characteristics. Three criteria of households' division were introduced. The households were divided according to the main source of maintenance, according to number of members in the household, according to the economic situation. As a result of these steps we got 18 households subsets. For each subset the linear econometric models were estimated and verified. These models illustrate the Engel's Law. In the next step of analysis the classification rules were generated with use of data mining techniques. Data mining analysis was made for the same subsets as previously econometric models. As an output we got the rules with economic interpretation, allowing for complement of households' characteristics resulting from research based on econometric techniques.

9

Social Capital and Economic Development. Automated Generation of Research Hypotheses

100%

Wachol D. , Glowka A. , Szapiro. T.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

141-165

EN

The relationship between social capital and economic development of European regions is investigated. Although relation under consideration appears not deterministic, the results remain in compliance with the theoretical conviction that countries with the high level of social capital are also characterized by the high level of production. Additionally, decision rules suggest the existence of relationship between quality of democracy and institutions and both - social capital and economic development. Variables describing the quality of government appeared in roots of the majority of the decision trees with social capital index as well as GDP.

10

Charakterystyka wybranych etapów procedury przygotowania danych do budowy modeli data mining

100%

Łapczyński M.

Zeszyty Naukowe Uniwersytetu Ekonomicznego w Krakowie

|

2012

|

nr 884

129-141

EN

The article examines the data preparation process that serves as the initial step in building data mining models. Following a short introduction to the stages of data pre-processing procedure, the main rules for creating derived variables are presented. The author describes ways of handling time variables and categorical variables and briefly looks at problems generated by the imprecise definition of variables, missing values and outliers.

11

Decision Rules Identification in Search for Attractive Investment

100%

Kochanowicz J. , Sojkowska E. , Szapiro T.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

119-140

EN

The problem of selection of attractive regions for investment was discussed. Based on empirical analysis, the need for analytical assistance for investors is advocated. Based on assumption on positive relation of social capital and growth, investigation focuses on social capital influence reflected in economic communication activities and intensity. There are considered mailing activities measured by number of circulating packages and indicators describing economic activities for regions. Then, AI tools are employed to generate rules which identify patterns in data. Results of investigation performed on real data describing Polish provinces and districts encourage to expand the approach and include electronic documents or billings of mobile telephones.

12

Knowledge-based information retrieval: The IKF project

100%

Dezsenyi C. e. a.

Tudomanyos es Muszaki Tajekoztatas ( Scientific and Technical Information; Hungarian journal of library and information science)

|

2004

|

tom 51

|

nr 5

193-208

EN

The Internet has become the leading information delivery medium in our days, yet it represents a huge, heterogeneous and distributed network from the point of view of information retrieval. Numerous software systems have appeared in the past years to support the automated information retrieval processes, applying useful information and knowledge retrieval techniques. Nevertheless, the core services of these systems do not provide real knowledge representation methods. The aim of our research is to design and develop a complete information and knowledge management system that applies integrated state-of-the-art knowledge intensive techniques. The research is part of the IKF project at the Department of Measurement and Information Systems of the Budapest University of Technology and Economics (BUTE) aims to analyze, design and implement a new intelligent knowledge-warehousing environment, allowing advanced knowledge management and decision support. The development is targeted towards specific financial application domain. This paper intends to provide a short review of knowledge-based information retrieval and extraction, and knowledge presentation technologies through the brief discussion of the project and the realized application. First, we present the architecture and the main features of the complete knowledge-based retrieval system. Then we focus on two major subjects: the document retrieval and information extraction system as well as the ontology-based knowledge management services. We also introduce some related topics briefly, such as structured information extraction with web wrappers, XML, and conceptualization with ontologies. Besides theory, we show some experimental results of the realized software systems.

13

Linguistic Summaries of Time Series as Tools for the Analysis of Trends

100%

Kacprzyk J. , Wilblik A. , Zadrozny S.

Roczniki Kolegium Analiz Ekonomicznych

|

2007

|

tom 17

166-186

EN

An approach to a linguistic summarization of numeric time series data is presenteded. Basically, the proposed summaries of time series refer to the linguistic summaries of what happens - in the sense of dynamics - to trends identified here with straight line segments of a piece-wise linear approximation of time series. Main attributes which are used for the summarization are: the duration, dynamics (slope) and variability of trends. The derivation of a linguistic summary of a time series is then related to a linguistic quantifier driven aggregation of trends. For this purpose the classic Zadeh's calculus of linguistically quantified propositions in its basic form is employed. An application to the absolute performance type analysis of time series data on daily quotations of an investment fund over an eight year period is presented as well as some interesting linguistic summaries obtained.

14

KOMPARÁCIA METÓD IDENTIFIKÁCIE DETERMINANTOV ŽIVOTNEJ ÚROVNE

88%

Birčiaková N. , Stávková J. , Trenz O.

Ekonomický časopis (Journal of Economics)

|

2016

|

tom 64

|

nr 6

560 – 574

EN

The article deals with the identification of factors of the standard of living. Ambiguous definition of the term “standard of living” requires for its quantification sufficient theoretical knowledge from the methodology field. The principal component analysis (PCA) was used to reduce the extensive amount of factors of the standard of living as well as for the determining of the most important ones. Data mining was used to compare the use of the PCA with various classification algorithms and later on it was tested with developments of the feature selection. The methods were applied on the sample of 2 783 respondents from 5 EU countries which represents areas of cultural similarities. The result of this effort is reduction from 99 considered factors of standard of living to final 45. Data mining helped to exclude 30 attributes and thus the final amount is set to 69. In case of comparison of these methods and their results it seems more appropriate the PCA over the feature selection method.

15

Evolution of Digital Video Consumption Patterns

88%

KUCHTA M. , MIKLOŠÍK A.

Communication Today

|

2017

|

tom 8

|

nr 2

58–69

EN

In today’s markets, companies have greatly shifted their focus towards customers as the entities creating the demand for companies’ products and sources of the revenues. The consumers shape and influence the current market trends with their desires and preferences; this requires reassessing and reshaping the offers and communications from companies. Most successful offer adjustments are built on relevant research and formed by consumer insights, often based on patterns of content consumption. The main aim of this article is to identify digital video marketing trends by combining two essential pillars: the data management possibilities related to the digital environment and the evolution of media consumption habits of consumers. The authors examine secondary data which is precisely selected, collected and shaped according to the research design requirements from available Consumer Barometer Research provided and processed by Google. Identification of the six most significant digital video trends to shape digital marketing in the following years can be considered the core of the authors’ research. The results are centred on innovation progress and consumers’ consumption habits and aim to outline possible approaches to digital video production and delivery to target groups for several future years.

16

THE USAGE OF MARGIN-BASED FEATURE SELECTION ALGORITHM IN IVF ICSI/ET DATA ANALYSIS

75%

Milewski R. , Malinowski P. , Milewska A. J. , Ziniewicz P.

Studies in Logic, Grammar and Rhetoric

|

2010

|

nr 21(34)

35-46

EN

In the case of infertility treatment, successful classification will facilitate understanding of various factors affecting the success of the process. Classification itself is an important data mining problem. Many classifications and constructions of the classifier algorithms are not able to cope with the analysis of the huge amount of factors associated with this process. Feature selection allows to significantly reduce the volume of analyzed data, while maintaining the classifier prediction quality. This leads to the rejection of nonessential measurements and time reductions.