Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies
This contribution tries to find an efficient way of a company´s property optimization. It searches for such a property structure that would ensure adequate benefit, respectively, the appreciation of own capital provided for remuneration. To carry out the calculation balance sheets, respectively thei...
Збережено в:
| Опубліковано в: : | Математичне моделювання в економіці |
|---|---|
| Дата: | 2016 |
| Автори: | , |
| Формат: | Стаття |
| Мова: | English |
| Опубліковано: |
Інститут телекомунікацій і глобального інформаційного простору НАН України
2016
|
| Теми: | |
| Онлайн доступ: | https://nasplib.isofts.kiev.ua/handle/123456789/131860 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Цитувати: | Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies / Z. Rowland, J. Vrbka // Математичне моделювання в економіці. — 2016. — № 3-4(7). — С. 36-42. — Бібліогр.: 20 назв. — англ. |
Репозитарії
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
nasplib_isofts_kiev_ua-123456789-131860 |
|---|---|
| record_format |
dspace |
| spelling |
Rowland, Z. Vrbka, J. 2018-04-04T19:36:14Z 2018-04-04T19:36:14Z 2016 Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies / Z. Rowland, J. Vrbka // Математичне моделювання в економіці. — 2016. — № 3-4(7). — С. 36-42. — Бібліогр.: 20 назв. — англ. 2409-8876 https://nasplib.isofts.kiev.ua/handle/123456789/131860 004.942 This contribution tries to find an efficient way of a company´s property optimization. It searches for such a property structure that would ensure adequate benefit, respectively, the appreciation of own capital provided for remuneration. To carry out the calculation balance sheets, respectively their parts informing about the company´s property are used, as well as income statements – the total taxed profit of all companies running their business in the CZ from 2006 to 2015. To find the model artificial neural networks are used – specifically a multi-layer perceptron network and a neural network of a radial basic function. The result is a neural structure that will help the building company find a suitable property structure, so that the company reaches the required ROE of 10% (a company is considered successful, if it reaches 10% and more on Return on Equity). The model is useful not only in company management but also in evaluating its performance and health by competitors, creditors or suppliers. Стаття присвячена аналізу одного з варіантів можливої оптимізації власного капіталу компанії. Наведено підхід до пошуку оптимальної структури капіталу, який дозволить забезпечити адекватну вигоду і зробити оцінку власного капіталу. Підхід базується на традиційному аналізі балансів, деталізації майна компанії, звітах про прибутки та збитки – загальних звітах для всіх компаній Чеської Республіки з 2006 по 2015 р. Для побудови моделі на основі нейронної мережі використовуються багатошарові мережі персептрона і нейронні мережі з радіально-базисною функцією. У результаті отримана нейронна структура для оптимізації капіталу будівельної компанії з необхідною рентабельністю власного капіталу в 10% (компанія вважається успішною, якщо вона досягає 10% і більше рентабельності власного капіталу). Модель призначена не тільки для управління компаніями, але й для оцінки їх продуктивності та працездатності конкурентами, кредиторами або постачальниками. Данная работа посвящена анализу одного из вариантов возможной оптимизации собственного капитала компании. Представлен подход к поиску оптимальной структуры капитала, которая позволит обеспечить адекватную выгоду и произвести оценку собственного капитала. Подход базируется на традиционном анализе балансов, детализации имущества компании, отчетах о прибылях и убытках – общих отчетах для всех компаний Чешской Республики с 2006 по 2015 г. Для построения модели на основе нейронной сети используются многослойные сети персептрона и нейронные сети с радиально-базисной функцией. В результате получена нейронная структура для оптимизации капитала строительной компании с необходимой рентабельностью собственного капитала в 10% (компания считается успешной, если она достигает 10% и более по рентабельности собственного капитала). Модель предназначена не только для управления компаниями, но и для оценки их производительности и работоспособности конкурентами, кредиторами или поставщиками. en Інститут телекомунікацій і глобального інформаційного простору НАН України Математичне моделювання в економіці Інформаційні технології в економіці Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies Оптимізація структури капіталу компанії з метою максимізації прибутку з використанням нейронних мереж на прикладі низки будівельних компаній Оптимизация структуры капитала компании с целью максимизации прибыли с использованием нейронных сетей на примере ряда строительных компаний Article published earlier |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| title |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies |
| spellingShingle |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies Rowland, Z. Vrbka, J. Інформаційні технології в економіці |
| title_short |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies |
| title_full |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies |
| title_fullStr |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies |
| title_full_unstemmed |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies |
| title_sort |
optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies |
| author |
Rowland, Z. Vrbka, J. |
| author_facet |
Rowland, Z. Vrbka, J. |
| topic |
Інформаційні технології в економіці |
| topic_facet |
Інформаційні технології в економіці |
| publishDate |
2016 |
| language |
English |
| container_title |
Математичне моделювання в економіці |
| publisher |
Інститут телекомунікацій і глобального інформаційного простору НАН України |
| format |
Article |
| title_alt |
Оптимізація структури капіталу компанії з метою максимізації прибутку з використанням нейронних мереж на прикладі низки будівельних компаній Оптимизация структуры капитала компании с целью максимизации прибыли с использованием нейронных сетей на примере ряда строительных компаний |
| description |
This contribution tries to find an efficient way of a company´s property optimization. It searches for such a property structure that would ensure adequate benefit, respectively, the appreciation of own capital provided for remuneration. To carry out the calculation balance sheets, respectively their parts informing about the company´s property are used, as well as income statements – the total taxed profit of all companies running their business in the CZ from 2006 to 2015. To find the model artificial neural networks are used – specifically a multi-layer perceptron network and a neural network of a radial basic function. The result is a neural structure that will help the building company find a suitable property structure, so that the company reaches the required ROE of 10% (a company is considered successful, if it reaches 10% and more on Return on Equity). The model is useful not only in company management but also in evaluating its performance and health by competitors, creditors or suppliers.
Стаття присвячена аналізу одного з варіантів можливої оптимізації власного капіталу компанії. Наведено підхід до пошуку оптимальної структури капіталу, який дозволить забезпечити адекватну вигоду і зробити оцінку власного капіталу. Підхід базується на традиційному аналізі балансів, деталізації майна компанії, звітах про прибутки та збитки – загальних звітах для всіх компаній Чеської Республіки з 2006 по 2015 р. Для побудови моделі на основі нейронної мережі використовуються багатошарові мережі персептрона і нейронні мережі з радіально-базисною функцією. У результаті отримана нейронна структура для оптимізації капіталу будівельної компанії з необхідною рентабельністю власного капіталу в 10% (компанія вважається успішною, якщо вона досягає 10% і більше рентабельності власного капіталу). Модель призначена не тільки для управління компаніями, але й для оцінки їх продуктивності та працездатності конкурентами, кредиторами або постачальниками.
Данная работа посвящена анализу одного из вариантов возможной оптимизации собственного капитала компании. Представлен подход к поиску оптимальной структуры капитала, которая позволит обеспечить адекватную выгоду и произвести оценку собственного капитала. Подход базируется на традиционном анализе балансов, детализации имущества компании, отчетах о прибылях и убытках – общих отчетах для всех компаний Чешской Республики с 2006 по 2015 г. Для построения модели на основе нейронной сети используются многослойные сети персептрона и нейронные сети с радиально-базисной функцией. В результате получена нейронная структура для оптимизации капитала строительной компании с необходимой рентабельностью собственного капитала в 10% (компания считается успешной, если она достигает 10% и более по рентабельности собственного капитала). Модель предназначена не только для управления компаниями, но и для оценки их производительности и работоспособности конкурентами, кредиторами или поставщиками.
|
| issn |
2409-8876 |
| url |
https://nasplib.isofts.kiev.ua/handle/123456789/131860 |
| citation_txt |
Optimization of a company’s property structure aiming at maximization of its profit using neural networks with the example of a set of construction companies / Z. Rowland, J. Vrbka // Математичне моделювання в економіці. — 2016. — № 3-4(7). — С. 36-42. — Бібліогр.: 20 назв. — англ. |
| work_keys_str_mv |
AT rowlandz optimizationofacompanyspropertystructureaimingatmaximizationofitsprofitusingneuralnetworkswiththeexampleofasetofconstructioncompanies AT vrbkaj optimizationofacompanyspropertystructureaimingatmaximizationofitsprofitusingneuralnetworkswiththeexampleofasetofconstructioncompanies AT rowlandz optimízacíâstrukturikapítalukompaníízmetoûmaksimízacíípributkuzvikoristannâmneironnihmerežnaprikladínizkibudívelʹnihkompaníi AT vrbkaj optimízacíâstrukturikapítalukompaníízmetoûmaksimízacíípributkuzvikoristannâmneironnihmerežnaprikladínizkibudívelʹnihkompaníi AT rowlandz optimizaciâstrukturykapitalakompaniiscelʹûmaksimizaciipribylisispolʹzovaniemneironnyhseteinaprimererâdastroitelʹnyhkompanii AT vrbkaj optimizaciâstrukturykapitalakompaniiscelʹûmaksimizaciipribylisispolʹzovaniemneironnyhseteinaprimererâdastroitelʹnyhkompanii |
| first_indexed |
2025-11-25T23:28:37Z |
| last_indexed |
2025-11-25T23:28:37Z |
| _version_ |
1850581249786118144 |
| fulltext |
~ 36 ~
Математичне моделювання в економіці, №3-4, 2016
UDK 004.942
Z. ROWLAND, J. VRBKA
OPTIMIZATION OF A COMPANY´S PROPERTY STRUCTURE AIMING
AT MAXIMIZATION OF ITS PROFIT USING NEURAL NETWORKS
WITH THE EXAMPLE OF A SET OF CONSTRUCTION COMPANIES
Abstract. This contribution tries to find an efficient way of a company´s
property optimization. It searches for such a property structure that would
ensure adequate benefit, respectively, the appreciation of own capital
provided for remuneration. To carry out the calculation balance sheets,
respectively their parts informing about the company´s property are used,
as well as income statements – the total taxed profit of all companies
running their business in the CZ from 2006 to 2015. To find the model
artificial neural networks are used – specifically a multi-layer perceptron
network and a neural network of a radial basic function. The result is a
neural structure that will help the building company find a suitable
property structure, so that the company reaches the required ROE of 10%
(a company is considered successful, if it reaches 10% and more on Return
on Equity). The model is useful not only in company management but also
in evaluating its performance and health by competitors, creditors or
suppliers.
Key words: property, ROE, artificial neural networks, model.
Introduction
The branch of construction belongs to the most demanding business branches in
general. In the article [1] characterizes building companies, as specific companies
dealing with building production. Especially in comparison to the industry, but also
to another branch, the building industry is characterized by a wide range of
specifics some of which may negatively influence its economy [2]. According to
[3] the difference of building companies from others is seen mainly in the
following: in the individual character of the product, organization of production
process, length of production cycle, mobility of the producer and stationarity of the
product and lower capacity use. In [4] adds the composition of property structure
among other differences or characteristic features. That should be chosen
appropriately to reach the long-term maximization of profit.
In the article [5] states that a company´s property structure is possible to be
understood from the balance sheet and is composed of two aggregated components,
in other words, fixed assets and current assets. According to [6] it is typical for
building companies to possess a significant share of current assets while the
greatest item of building production is made of stock, without a doubt. Due to this
fact it is very important to control their correct management. In case the company
binds more than needed it acts inefficient [7].
Ó Z. Rowland, J. Vrbka, 2016
~ 37 ~
Математичне моделювання в економіці, №3-4, 2016
The main aim of a building company´s origin and pursuit is, as in any other
company´s case, is reaching the highest efficiency of input capital possible. That
must be ensured especially by the company prosperity, which will be reached by
building company profitability [6]. Companies are founded exactly in order to
generate profit. One of the most important indicators for the measuring of a
building company´s performance (profit), which is most interesting for managers,
shareholders and investors, is the indicator of Return on Equity (ROE). ROE
basically represents the measure of how efficiently the company uses the
shareholders´ means to consequently generate profit [8]. In [9] claims that through
ROE it is possible to determine whether the company is profit creator or whether,
on the contrary, it does not generate profit.
ROE indicator will be found through the following relation [10]:
(1)
According to [11] Return on Equity rate may be considered positive in case its
value is higher than 12%. For the purpose of this article a company will be
considered successful if it reaches 10% and more of ROE.
In today´s modern world the number of companies, using ANNs (Artificial
Neural Networks) to find a suitable and efficient way of property structure
optimization [12]. In [13] claims that it was possible to meet these systems,
computer software working on the basis of special mathematic algorithms,
respectively, for the first time in the area of neurology. According to [14] ANNs
are characterized especially by their high ability to analyse a huge volume of
information. Other abilities of artificial neural networks, according to [15] include
the abilities to learn, to generalize data, to memorize, to produce new information,
etc.
The MPL Multi-Layer Perceptron Neural Network and RBF – Radial Basic
Network may be classified among the most frequently used neural networks [16].
These two types of ANNs will be used throughout this article to find a model
which would optimize a company´s property efficiently. In [17] characterize MLP
as a forward artificial neural network which consists of at least two layers of
perceptrons (neurons). In each layer all inputs of individual neurons are connected
to the outputs of the previous layer. The outputs head towards the following layer
only [18]. According to [17] MLP is a modification of a classic linear perceptron
and it is able to differentiate information impossible to be separated linearly. The
RBF neural network has a similar structure to the previous networks. That one is,
according to [19] composed of two layers of forward networks and is used mainly
for approximation function and a time line of prognoses, for classification and
clustering of tasks (interpolation, time line modelling, speech distinction, 3D
modelling, data fusion).
The aim of this contribution is to find an efficient way of a company´s property
optimization. Thus, such a property structure that will ensure an adequate profit,
with the example of building companies in the Czech Republic.
1. Material and methodology
For the purpose of analyzis the data of building companies in the Czech Republic,
from 2006 to 2015, has been chosen. Specifically, they are companies classified in
~ 38 ~
Математичне моделювання в економіці, №3-4, 2016
the Building F section, i.e. [20]: ´This section includes specialized as well as non-
specialized building activities. They include new-building works, repairs, building
extensions, reconstructions, engineering works, building of pre-fabricated objects
in the construction-site, and temporary buildings.
These include the building of complete flat, office and shopping-mall
buildings, other public buildings, agriculture buildings, sports halls and
gymnasiums, etc., on the one hand, and the building of highways, roads, bridges,
tunnels, railways, runways, ports and other aquatic constructions, irrigation canals,
sewerage, industrial objects, conduits and power lines, open sports stadiums and
playgrounds, etc., on the other hand.´ We were interested in the data of the
companies, such as identification number, name of the given company, the amount
of current assets in thousands of CZK, the amount of fixed assets in CZK, profit or
loss of the given company.
The data file will thus create a table where the line will represent the company
and a specific year of economy. In total, the file contains 66 743 of record lines.
Records of companies in disposal and records of companies terminating their
activity in the given year (potentially extreme values) were removed from the file.
Our aim is basically finding the production curve that uses two inputs – current
assets and fixed assets. At the same time, we are looking for such a combination of
inputs that will bring the most optimal profit (respectively production) to the
building company.
To prepare the data file MS Excel will be used. For the purpose of the
calculation the DELL Statistica software in version No. 7 and 12 will be used.
Consequently, it will be processed through automated neural networks.
Used variables are continuous. Thus, we will use a module of time series used
through regression. The data will be divided into three groups:
– Training: 70%,
– Testing: 15%,
– Validation: 15%.
The seed for a random selection was determined at a value of 1000.
Downsampling will be run randomly.
Consequently, 1000 random artificial neural structures will be generated, out of
which 5 most appropriate results will be preserved1.
These kinds of neural networks will be used:
1. A neural network of Radial Basic Function (further as RBF),
2. A multi-layer perceptron neural network – three-layer (further on as MLP).
We will use the following: linear function, step function, saturating linear
function, sigmoid function and hyperbolic tangent function as an activating
function in a hidden and output layer of neurons.
In the hidden RBF layer up to 9 neurons will be used. In the hidden layer of the
three-layer perceptron network up to 20 neurons will be used. Other settings will be
default.
1 We will determine this through the smallest squares method. If the differences between
newly generated networks will be no more significant the training session will be ended.
~ 39 ~
Математичне моделювання в економіці, №3-4, 2016
2. Results and Discussion
The data was divided into three sets – training, testing and validation. The choice
was carried out randomly. Based on the data sets 1000 neural networks were
generated. Five of them were preserved – those with the best statistics. An
overview of the obtained and preserved neural networks is given in Table No. 1.
Table 1 – An overview of generated and preserved neural networks
In Name of
Network
Train.
Performance
Test.
Performance
Valid.
Performance Train. Fault Testing
Fault
1 MLP 2-8-1 0,664731 0,618323 0,766145 193071859 104324593
2 MLP 2-6-1 0,667188 0,620293 0,766498 191987650 103794019
3 MLP 2-9-1 0,664684 0,620674 0,765502 193079223 103764588
4 MLP 2-7-1 0,664716 0,619914 0,765698 193104285 103932935
5 MLP 2-17-1 0,667793 0,621257 0,766153 191653012 103637452
In Name of
Network
Validation
Fault Train. Algorithm Fault
Function
Activ.
Hidden
Layer
Output
Act.
Layer
1 MLP 2-8-1 74849911 BFGS (Quasi-Newton) 9 Sum of
Squares Logistic Tan
2 MLP 2-6-1 74658406 BFGS (Quasi-Newton) 18 Sum of
Squares Tan Tan
3 MLP 2-9-1 74901561 BFGS (Quasi-Newton) 10 Sum of
Squares Sinus Logistic
4 MLP 2-7-1 74873824 BFGS (Quasi-Newton) 8 Sum of
Squares Sinus Logistic
5 MLP 2-17-1 74741476 BFGS (Quasi-Newton) 15 Sum of
Squares Tan Logistic
Source: Authors
It is interesting that all preserved neural networks are based on a multi-layer
perceptron neural network. The table proves that all exhibit similar characteristics
with regard to performance and fault, in all three data sets. Individual networks
only differ in the activating networks used in the hidden layer of neurons and in the
output layer of neurons. In the hidden layer of neurons, they use a logistic function,
a hyperbolic tan and the sinus function. Hyperbolic tan and logistic function, are
then used in the hidden neural layer. Thus, at the first sight we could generalize the
fact that MLP is suitable for regression tasks.
Further, correlation coefficients of all generated and preserved networks were
analysed, having been divided into training, testing and validation set of data. It
may be deduced from the results that in all three sets of data there exists a
correlation. Yet it is not so high. It moves throughout sets and individual networks
from almost 0.62 to almost 0.77. A comparison descriptive characteristics among
individual data sets follows. It is clear that partial results are relatively similar.
Although they do differ in some items, the differences are not significant.
In Figure No. 1, to get a better idea, we may observe the distribution of current
asset combinations, the distribution of fixed assets and building companies´ profit
in the CZ, predicted by the MLP 2-8-1 network. Similarly, the distribution of
~ 40 ~
Математичне моделювання в економіці, №3-4, 2016
current assets´ combinations was created, as well as the distribution of fixed assets
and the profit of building companies in the CZ, predicted by the rest of the chosen
neural networks.
Source: Authors
Figure 1 – Combination of Current Assets, Fixed Assets and Profit of building
companies in the CZ according to the MLP 2-8-1
Prediction differs from the reality only in lower values of current assets and in
lower values of fixed assets. In higher values prediction reaches reality very
closely. Models even respect the deviated value, when the company generated
profit at a low volume of current assets and at a high volume of fixed assets. It may
be assumed that if the data was divided into four quadrants:
1. A high volume of current assets and at the same moment a high volume of
fixed assets,
2. A high volume of current assets and at the same moment a low volume of
fixed assets,
3. A low volume of current assets and at the same moment a high volume of
fixed assets,
4. A low volume of current assets and at the same moment a low volume of
fixed assets.
We would find out that the highest level of correlation may be reached in
quadrant No.1. It would be followed by quadrants no. two and three. The fourth
quadrant would be correlated the least.
To verify this idea, we created a sensitivity analysis of economy result based on
partial input variables. Profit is based on both input variables. It depends much
~ 41 ~
Математичне моделювання в економіці, №3-4, 2016
more on the volume of current assets where the values move around the interval of
more than 1.641 to almost 1.675 thousand of CZK. In case of fixed assets, the
values move around almost 1.007 up to almost 1.027 thousand CZK. The result
also reflects the current structure of building companies´ property, where a huge
volume is made of fixed assets.
Conclusion
The aim of this contribution has been to find an efficient way of optimization of a
company´s property. Thus to find such a property structure that would ensure an
adequate profit based on the example of building companies in the Czech Republic.
The aim of the contribution has been fulfilled. 1000 neural structures were
generated out of which five with the best characteristics were preserved. With
regard to a not-entirely-satisfactory correlation among the input and output variable
a closer analysis of interdependence of individual variables had to be carried out.
Thanks to that it turned out that the generated and preserved models were much
more exact if they predict the results of large enterprises, i.e. enterprises with a
large volume of current and fixed assets. Moreover, the analysis clearly proves that
the company´s economic result is much more sensitive towards changes in current
assets volume than towards the changes of fixed assets volume. Thus a clear
recommendation towards capital goods consumption management is given –
towards the company´s cost policy. Building enterprises have to strive to be as
efficient as possible working with material in construction sites. Fixed assets
consumption is fixed.
The results offer a clear direction for the purpose of another paper – to divide
building companies into smaller groups according to their size. Further, it would be
suitable to pay attention to partial property items.
Models as such are not very useful in practice at the evaluated detail rate. Yet
they do allow us to guess, having carried out the above-mentioned changes, a large
potential for enterprise management (especially in case of large enterprises).
REFERENCES
1. Hoffmann, S., Wicke, D., & Cadez, I. (2015). Survey of sustainability in construction
companies. Bautechnik, 92(10), pp. 725-729.
2. Vochozka, M. & Stehel, V. (2012). Regionální disparity ve vývoji stavebního odvětví v
České republice. [Regional disparities in the development of the construction sector in the
Czech Republic]. Littera Scripta, 5(2), pp. 187-202.
3. Horta, I., & Camanho, A. (2013). Company failure prediction in the construction
industry. Expert Systems with Applications, 16(40), pp. 6253-6257.
4. Lehutová, K., Križanová, A. & Klieštik, T. (2013). Quantification of Equity and Debt
Capital Costs in the Specific Conditions of Transport Enterprises, In: Proceedings of the
17th International Conference on Transport Means, Transport Means. Kaunas, Lithuania,
pp. 258-261.
5. Wong, W. P., Gholipour, H. F., & Bazrafshan, E. (2012). How efficient are real estate
and construction companies in Iran's close economy? International Journal of Strategic
Property Management, 16(4), pp. 392-413.
6. Yang, Z., Lin, L., & Li, H. (2010). Ownership Structure and Financial Reporting
Transparency: The Empirical Evidence from Construction and Real Estate Listed
~ 42 ~
Математичне моделювання в економіці, №3-4, 2016
Companies in China. Proceedings of the 2010 International Conference on Construction
and Real Estate Management. Brisbane, Australia, pp. 816-820.
7. Siskina, S., Juodis, A., & Apanaviciene, R. (2009). Evaluation of the competitiveness of
construction company overhead costs. Journal of Civil Engineering and
Management, 15(2), pp. 215-224.
8. de Wet, J. H. V. H., & du Toit, E. (2007). Return on equity: A popular, but flawed
measure of corporate financial performance. South African Journal of Business
Management, 38(1), pp. 56-69.
9. Kijewska, A. (2016). Determinants of the Return on Equity Ratio (Roe) on the Example
of Companies from Metallurgy and mining sector in Poland. Metalurgija, 55(2), pp. 285-
288.
10. Vojteková, M, Bartošová, V. (2009). Comparability and mathematical aspects of the
use of indicators of financial analysis in the evaluation of the company. Ekonomicko-
manažerské spektrum, 3(2), pp. 72-76.
11. Ichsani, S., & Suhardi I, A. R. (2015). The Effect of Return on Equity (ROE) and
Return on Investment (ROI) on Trading Volume. Procedia – Social and Behavioral
Sciences, 211, pp. 896-902.
12. Wilczyńska-Piliszek, A. J., Piliszek, S., & Falandysz, J. (2012). Use of quantitative-
structure property relationship (QSPR) and artificial neural network (ANN) based
approaches for estimating the octanol-water partition coefficients of the 209 chlorinated
trans -azobenzene congeners. Journal of Environmental Science and Health, Part B, 47(2),
pp. 111-128.
13. Izeboudjen, N., Larbes, C., & Farah, A. (2014). A new classification approach for
neural networks hardware: from standards chips to embedded systems on chip. Artificial
Intelligence Review, 41(4), pp. 491-534.
14. Mileris, R., & Boguslauskas, V. (2011). Credit Risk Estimation Model Development
Process: Main Steps and Model Improvement. Engineering Economics, 22(2), pp. 126-133.
15. Luzar, M., Sobolewski, Ł., Miczulski, W., Korbitcz, J., Díaz Lantanda, A., Munoz -
Guijosa, J. M., & Muňoz Sanz J. L. (2014). Prediction of corrections for the Polish time
scale UTC (PL) using artificial neural networks: from standards chips to embedded systems
on chip. Lubrication Science, 26(3), pp. 141-162.
16. Michal, P., Vagaská, A., Gombár, M., Kmec, J., Spišák, E., & Kučerka, D. (2015).
Usage of Neural Network to Predict Aluminium Oxide Layer Thickness. The Scientific
World Journal, 2015, pp. 1-10.
17. Kumar, M., & Yadav, N. (2011). Multilayer perceptrons and radial basis function
neural network methods for the solution of differential equations: A survey. Computers &
Mathematics with Applications, 62(10), pp. 3796-3811.
18. Piasecki, A., Jurasz, J., & Marszelewski, W. (2016). Application of Multilayer
Perceptron Artificial Neural Networks to Mid-Term Water Consumption Forecasting - A
Case Study. Ochrona Srodowiska, 38(2), pp. 17-22.
19. Pazouki, M., Wu, Z., Yang, Z., & Moeller, D. P. F. (2015). An Efficient Learning
Method for RBF Neural Networks. In Proceedings of the International Joint Conference on
Neural Networks. Killarney, Ireland. pp. 1-6.
20. CZ NACE, 2016. CZ NACE – F Stavebnictví. [F Construction]. [online]. [accessed:
2016-10-10]. Available from: http://www.nace.cz/nace/f-stavebnictvi/
Стаття надійшла до редакції 29.10.16.
Introduction
Introduction
1. Material and methodology
2. Results and Discussion
Conclusion
|