Wyniki wyszukiwania - Biblioteka Nauki

1

Clustering of European Countries with Respect to Food Consumption

100%

Dudek H. , Orłowski A. , Warsaw Agricultural University D. o. E. a. I. , Polish Academy of Science W. A. U. a. I. o. P.

Acta Universitatis Lodziensis. Folia Oeconomica

|

2006

|

tom 196

EN

Problem of clustering of European countries with respect to food consumption is considered. Data related to average yearly per capita consumption of 14 main categories of food products in 39 countries are collected and analysed. Food consumption data for two years: 2000 and 1993 are elaborated. The year 2000 was because there are no more recent data sets available. The year 1993 was chosen as a good reference point: data for that year are the oldest complete. To perform a reasonable grouping of countries the cluster analysis is performed. As a proper number of cluster is not known in advance, hierarchical methods offered by statistical packages Statgraphics are used. The desirable number of clusters is estimated by distance matrices analysis, dendrograms, and graphical representations of distance between clusters with respect to different clustering stages. Squared Euclidean distance is used as a measure of similarity. It is remarkable that all hierarchical methods applied in this paper, apart from nearest neighborhood approach, lead to very similar classification results. Therefore we believe that obtained results provide a valuable and objective insight into the problem of diversification of food consumption in Europe. It has been verified that in spite of visible changes in food consumption in investigated countries, sets of countries belonging to particular clusters obtained for 2000 and for 1993 are almost indistinguishable.

PL

W artykule rozważono zagadnienie pogrupowania państw europejskich ze względu na konsumpcję żywności. Zgromadzono dane o rocznym spożyciu na osobę 14 głównych grup produktów żywnościowych w 39 państwach. Dane dotyczą konsumpcji żywności w latach 2000 oraz 1993. W celu pogrupowania państw wykorzystano analizę skupień. Z uwagi na brak przesłanek dotyczących liczby skupień zastosowano hierarchiczne metody aglomeracyjne, oprogramowane w pakietach statystycznych Statgraphics. Liczbę skupień ustalono na podstawie analizy macierzy odległości, dendrogramów oraz wykresów odległości skupień względem etapów grupowania. Za miarę podobieństwa przyjęto kwadrat odległości euklidesowej. Ustalono, że poza metodą najbliższego sąsiedztwa, wszystkie hierarchiczne metody aglomeracyjne prowadzą do skupień o zbliżonym zestawie państw. Na podstawie wykonanej analizy skupień stwierdzono, że mimo zmian w spożyciu produktów żywnościowych w poszczególnych krajach, zestawy państw w otrzymanych skupieniach w roku 2000 i 1993 były niemal identyczne.

2

Trends in Cigarettes Consumption in Poland According to Expotential Smoothing and Autoregressive Models

80%

Jałowiecka E. , Jałowiecki P. , Orłowski A. , Chair of Informatics W. U. o. L. S.

Acta Universitatis Lodziensis. Folia Oeconomica

|

2009

|

tom 228

PL

Polski przemysł wyrobów tytoniowych przechodzi w ostatnich latach znaczące przemiany związane z akcesją Polski do Unii Europejskiej. Stanowi on ważny sektor polskiej gospodarki generując 7,5% dochodów budżetu państwa. W pracy porównano prognozy spożycia papierosów w latach 2006-2010 przygotowane w oparciu o wybrane modele wyrównywania wykładniczego oraz autoregresyjne na podstawie danych historycznych z lat 1995-2005. Główną uwagę skoncentrowano na trendzie w prognozach. Identyfikację modeli autoregresyjnych przeprowadzono przy użyciu metod typu „corner” oraz rozszerzonej funkcji autokorelacji. W celu zwiększenia wiarygodności, prognozy przygotowano z uwzględnieniem zidentyfikowanych wartości odstających. Uzyskane wyniki porównano z danymi szacunkowymi uzyskanymi z Głównego Urzędu Statystycznego oraz z wynikami prognoz uwzględniających jako dodatkową zmienną produkcję papierosów przygotowanymi z zastosowaniem techniki „prewhitening”. Przeprowadzono dyskusję zalet i wad zastosowanych metod.

EN

Polish tobacco industry has been recently changing significantly due to accession of Poland to EU. It is one of the prime sector of polish economy. It generates every year about 7% of budget incomes on average. The aim of this paper is to compare some forecast methods of cigarettes consumption in 2006-2010. The models used exponential smoothing and autoregression theory. The forecasts were estimated on historical data from 1995-2005. The main attention was focused on the trends in prediction. Identification, the most crucial stage in fitting autoregressive models exploited different approach such as the comer method and extended sample autocorrelations. The outlier selection techniques were also applied to get more reliable estimates. The results were compared to the predicted values obtained from Central Statistical Office and to the results of forecasts taking cigarettes production into consideration due to prewhitening technique. The advantages and drawbacks of different methods are discussed.

3

The Perspectives of Bayesian Methods for Modeling Financial Reserves in Insurance

80%

Karwański M. , Jałowiecki P. , Orłowski A. , Chair of Informatics W. U. o. L. S.

Acta Universitatis Lodziensis. Folia Oeconomica

|

2009

|

tom 228

PL

Celem niniejszej pracy było porównanie wyników szacowania rezerw metodami tradycyjnymi: model Macka, modele GLM z szacowaniem zmienności metodą bootstrap oraz metody Bayesa MCMC. Analizy przeprowadzone zostały na danych jednej z dużych amerykańskich firm ubezpieczeniowych. Takie porównanie, może być jednym z głosów w dyskusji, w kontekście trwającego aktualnie procesu tworzenia standardów rachunkowości ubezpieczeniowej oraz systemów monitorujących adekwatność rezerw.

EN

Building reserves for liabilities is an important issue in the financial statement o f a general insurance company. The purpose o f this paper is to present models for prediction IBNR (incurred but not reported) reserves. The modeling is based on data which describes the claims settlement o f a car insurance portfolio - the data consists o f about 60,000 claims, which incurred in 2001—2006. Several models o f the claims prediction are proposed, from estimation in traditional deterministic Chain Ladder and the Poisson GLM model (Kramer 1998) - commonly used techniques in practices - to Markov Chain Monte Carlo. The models presented show the significant differences in variance o f IBNR reserves. From that point o f view the Bayesian approach has some characteristics that make it particularly attractive for their use in actuarial practice.

4

SOME PROPOSAL OF THE TEST FOR A RANDOM WALK DETECTION AND ITS APPLICATION IN THE STOCK MARKET DATA ANALYSIS

80%

Dudziński M. , Furmańczyk K. , Orłowski A.

Metody Ilościowe w Badaniach Ekonomicznych

|

2018

|

tom 19

|

nr 4

339-346

EN

According to the numerous groups of theoreticians and practitioners, who act in the area of financial markets, changes in the stock prices are random and it is almost infeasible to predict them correctly using historical data. This approach is based on the random walk theory, which states that the price of financial instrument in the subsequent time point is the sum of its price in the previous time point and some random variable with a finite variance, i.e. it is modeled with the use of a stochastic process called a random walk. The random walk hypothesis stands in contradiction to the beliefs of the ordinary technical analysis followers, where the prediction is carried out on the grounds of existing trends, and furthermore, this hypothesis regards such a modeling of financial markets as incorrect. In our work, we construct statistical test for a random walk detection, which is based on the first arcsine law. We also present simulation results that allow to check the quality of the proposed test, as well as we show the application of the introduced test in the stock exchange data analysis.

5

BCT Boost Segmentation with U-net in TensorFlow

61%

Wieczorek G. , Antoniuk I. , Kurek J. , Świderski B. , Kruk M. , Pach J. , Orłowski A.

Machine Graphics and Vision

|

2019

|

tom Vol. 28, No. 1/4

25--34

EN

In this paper we present a new segmentation method meant for boost area that remains after removing the tumour using BCT (breast conserving therapy). The selected area is a region on which radiation treatment will later be made. Consequently, an inaccurate designation of this region can result in a treatment missing its target or focusing on healthy breast tissue that otherwise could be spared. Needless to say that exact indication of boost area is an extremely important aspect of the entire medical procedure, where a better definition can lead to optimizing of the coverage of the target volume and, in result, can save normal breast tissue. Precise definition of this area has a potential to both improve the local control of the disease and to ensure better cosmetic outcome for the patient. In our approach we use U-net along with Keras and TensorFlow systems to tailor a precise solution for the indication of the boost area. During the training process we utilize a set of CT images, where each of them came with a contour assigned by an expert. We wanted to achieve a segmentation result as close to given contour as possible. With a rather small initial data set we used data augmentation techniques to increase the number of training examples, while the final outcomes were evaluated according to their similarity to the ones produced by experts, by calculating the mean square error and the structural similarity index (SSIM).

6

Textural features based on run length encoding in the classification of furniture surfaces with the orange skin defect

51%

Pach J. , Chmielewski L. J. , Orłowski A. , Kruk M. , Kurek J. , Świderski B. , Antoniuk I. , Wieczorek G. , Śmietańska K. , Górski J.

Machine Graphics and Vision

|

2019

|

tom Vol. 28, No. 1/4

35--45

EN

Textural features based upon thresholding and run length encoding have been successfully applied to the problem of classification of the quality of lacquered surfaces in furniture exhibiting the surface defect known as orange skin. The set of features for one surface patch consists of 12 real numbers. The classifier used was the one nearest neighbour classifier without feature selection. The classification quality was tested on 808 images 300 by 300 pixels, made under controlled, close-to-tangential lighting, with three classes: good, acceptable and bad, in close to balanced numbers. The classification accuracy was not smaller than 98% when the tested surface was not rotated with respect to the training samples, 97% for rotations up to 20 degrees and 95.5% in the worst case for arbitrary rotations.

7

Data augmentation techniques for transfer learning improvement in drill wear classification using convolutional neural network

51%

Kurek J. , Antoniuk I. , Górski J. , Jegorowa A. , Świderski B. , Kruk M. , Wieczorek G. , Pach J. , Orłowski A. , Aleksiejuk-Gawron J.

Machine Graphics and Vision

|

2019

|

tom Vol. 28, No. 1/4

3--12

EN

This paper presents an improved method for recognizing the drill state on the basisof hole images drilled in a laminated chipboard, using convolutional neural network (CNN) and data augmentation techniques. Three classes were used to describe the drill state: red - for drill that is worn out and should be replaced, yellow - for state in which the system should send a warning to the operator, indicating that this element should be checked manually, and green - denoting the drill that is still in good condition, which allows for further use in the production process. The presented method combines the advantages of transfer learning and data augmentation methods to improve the accuracy of the received evaluations. In contrast to the classical deep learning methods, transfer learning requires much smaller training data sets to achieve acceptable results. At the same time, data augmentation customized for drill wear recognition makes it possible to expand the original dataset and to improve the overall accuracy. The experiments performed have confirmed the suitability of the presented approach to accurate class recognition in the given problem, even while using a small original dataset.

8

Context-based segmentation of the longissimus muscle in beef with a deep neural network

51%

Talacha K. , Świderski B. , Kurek J. , Kruk M. , Półtorak A. , Chmielewski L. J. , Wieczorek G. , Antoniuk I. , Pach J. , Orłowski A.

Machine Graphics and Vision

|

2019

|

tom Vol. 28, No. 1/4

47--57

EN

The problem of segmenting the cross-section through the longissimus muscle in beef carcasses with computer vision methods was investigated. The available data were 111 images of cross-sections coming from 28 cows (typically four images per cow). Training data were the pixels of the muscles, marked manually. The AlexNet deep convolutional neural network was used as the classifier, and single pixels were the classified objects. Each pixel was presented to the network together with its small circular neighbourhood, and with its context represented by the further neighbourhood, darkened by halving the image intensity. The average classification accuracy was 96%. The accuracy without darkening the context was found to be smaller, with a small but statistically significant difference. The segmentation of the longissimus muscle is the introductory stage for the next steps of assessing the quality of beef for the alimentary purposes.

9

Classifiers ensemble of transfer learning for improved drill wear classification using convolutional neural network

51%

Kurek J. , Antoniuk I. , Górski J. , Jegorowa A. , Świderski B. , Kruk M. , Wieczorek G. , Pach J. , Orłowski A. , Aleksiejuk-Gawron J.

Machine Graphics and Vision

|

2019

|

tom Vol. 28, No. 1/4

13--23

EN

In this paper we introduce the enhanced drill wear recognition method, based on classifiers ensemble, obtained using transfer learning and data augmentation methods. Red, green and yellow classes are used to describe the current drill state. The first one corresponds to the case when drill should be immediately replaced. The second one denotes a tool that is still in a good condition. The final class refers to the case when a drill is suspected of being worn out, and a human expert evaluation would be required. The proposed algorithm uses three different, pretrained network models and adjusts them to the drill wear classification problem. To ensure satisfactory results, each of the methods used was required to achieve accuracy above 90% for the given classification task. Final evaluation is achieved by voting of all three classifiers. Since the initial data set was small (242 instances), the data augmentation method was used to artificially increase the total number of drill hole images. The experiments performed confirmed that the presented approach can achieve high accuracy, even with such a limited set of training data.