Software Solution for the Implementation of a Predictive Analytics
System for Investment Instruments
NATALIA MAMEDOVA1, OLGA STAROVEROVA1, ALEXEY EPIFANOV1,
HUAMING ZHANG2, ARKADIY URINTSOV1
1Basic Department of digital economy, Higher School of Cyber Technologies,
Mathematics and Statistics, Plekhanov Russian University of Economics, RUSSIA
2Deputy Dean, School of Economics, Shanxi University of Finance and Economics, CHINA
Abstract: This article raises the issues of research investment support tools and the study of existing IT
solutions in the field of predictive analytics investment solutions. The research request is based on the lack of
accuracy, and objectivity of existing methods of investment analysis and means of its automation. A review of
existing technical solutions and technologies is carried out. The process of analyzing investment instruments
has been studied, and bottlenecks in existing approaches to analysis have been identified. A solution for
implementing a system of predictive analytics of investment instruments has been developed. The solution is
based on the business requirements and functional requirements of the software development company.
Key-Words: predictive analytics, investment tool, investment analytics, information system, toolkit.
Received: September 23, 2022. Revised: December 4, 2022. Accepted: January 7, 2023. Published: February 14, 2023.
1 Introduction
It is believed that predictive analytics, as an end-to-
end technology, has great potential, including in the
financial and credit sphere, [1], [2]. There are two
approaches in the field of financial market analysis -
technical analysis and fundamental analysis. Each of
them has its limitations. Fundamental analysis, due
to the expert assessment used in its basis, is
subjective and limited due to the presence of a
human factor. Technical analysis has a high level of
accuracy, but it becomes hostage to the prediction
algorithms embedded in it and the ratio of key
indicators. The ideal solution seems to be the
implementation of predictive analytics solutions that
will eliminate the bottlenecks of existing
approaches. That is, they will increase the
objectivity and accuracy of forecasts by using
internal and external historical data, as well as real-
time data. For such a segment of the financial
market as investing, both the above approaches and
their limitations are fair.
The development of a predictive analytics system
for investment instruments is a complex process.
The most responsible and difficult stage of the
process is the selection of the source data and
working with the data, that is, the scope of the Data
Scientist technology.
The purpose of the research was to develop a
tool for implementing a system of predictive
analytics of investment instruments. The study was
conducted in many areas. Firstly, the results of
research on the topic of predictive analytics were
summarized. Secondly, to identify the technological
basis of the subject area, existing IT solutions
supporting predictive analytics and specializing in
the investment segment were studied. Based on this
information, data on the bottlenecks of such IT
solutions were summarized. The development of a
tool for implementing a predictive analytics system
was also preceded by an analysis of the mechanism
implemented by analysts to predict changes in the
investment market. The research process described
in the article and its results have an implementation
potential since the processes of implementing the
predictive analytics information system are of
commercial interest to information system
developers.
The following parameters became the functional
boundaries of the development toolkit:
• The user should be able to view the forecast at
the current time, without being able to view the
forecast history;
• There are no notifications about forecast
changes (to reduce the volatility of decisions
made based on the forecast);
• Historical data is provided by information
resources located on the territory of the Russian
Federation;
• The activities of the organization of the investor
company and the consultant company are
carried out on the territory of the Russian
Federation.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
18
Volume 18, 2023
2 Problem Formulation
Predictive analytics (other names – advanced
analytics, advanced qualitative analysis, intelligent
analysis) combines a variety of forecasting methods.
It is aimed at preventing economic losses and
finding the optimal logic of actions when making
decisions, [3]. The analyst, using the methods of
predictive analytics, looks for patterns in historical
and transactional data and forms an educational
program on potential risks and opportunities. A
forecast is made based on several predictor variables
in a given data set. Then the aggregated data is
received by the decision-maker.
Predictive analytics is perfectly consistent with
the SMART concept of goal setting and with the
concept of Data-Driven Decision, [4]. And the
benefits of using predictive analytics for business
are considered obvious. Advanced quantitative
analysis has demonstrated advantages in various
sectors of the economy, [1], [5]. Solutions based on
it allow you to find opportunities, and reduce
uncertainty in risk management – all these are
effects that reduce the impact of negative business
factors. The results of predictive analytics improve
the quality of business planning. As a final result,
the detection of significant patterns, forecasting, and
timely response to changes increase the
competitiveness of an economic agent.
Traditionally, the following stages of predictive
analytics implementation are distinguished: data
connection; data preparation, analysis, and
visualization; development of alternatives and
testing of data models; application of predictive
models; evaluation and/or forecasts of future results,
[6].
Working out alternatives and testing data models
is one of the most time-consuming stages. Its results
can either determine the quality of the next stage
(meaning the use of predictive models) or "reset"
the stages already passed and return the analysis to
its original position, starting the data connection
stage again.
The development of alternatives and testing of
data models is carried out in two directions –
supervised learning and unsupervised learning. A
detailed description of the order and features of the
course of learning processes is not mandatory for
the topic of our study, so we will limit ourselves to a
summary of the essence based on the results of
summarizing data from thematic sources, [7], [8].
Supervised learning is divided into two large
categories: regression for quantitative responses
(numerical value) and classification, which uses
categorical variables of the response. Unsupervised
learning is used to derive conclusions from datasets
consisting of input data without specifying
responses. The most common method of
unsupervised learning is cluster analysis, which is
used to investigate and search for hidden patterns in
data.
Today you can find three of the most popular
variations of the implementation of predictive
analytics of investment instruments: technical
analysis; trading advisor; predictive analytics
systems.
The basic principle of technical analysis is to
analyze the indicators of indicators. As a result,
advice is taken to buy, sell or wait for investment
actions. The scope of application of this type of
predictive analytics implementation is mainly
specialized sites dedicated to financial trading tools.
If we talk about the specifics of the forecast, then it
is completely determined by the price movement in
the investment asset market. And in case of a strong
price shift, the forecast itself may change to the
opposite.
The implementation of predictive analytics by
the type of trading adviser includes technical
analysis itself but works with combinations of
indicators and with their specific settings. The result
of the work is a signal for the investor, who advises
on buying or selling an investment asset. The
Trading Advisor has a greater forecasting accuracy
compared to technical analysis, as they are subjected
to more fine-tuning in the investment strategy
testing mode.
The latter type of implementation uses more
complex predictive analytics algorithms. Such
systems use machine learning models to train the
system and build a forecast based on the results of
collecting and analyzing data on changes in the
price of an asset and based on studying the
indicators of technical analysis indicators. The result
of the predictive analytics system is a forecast for a
given time period, for example, whether the price
will continue to move in the current direction.
Experts estimate the accuracy of the forecast of this
type of predictive analytics as the highest – about
70% with an error of 2-3%. At the same time, it
should be noted that the system does not offer a
forecast of price changes in percentage or
quantitative value.
Let's present the algorithm of the predictive
analysis system using the example of SAP
Predictive Analytics:
1. Selection of the data source(s).
2. Uploading quotes for a given time interval of
trading platforms and a given calendar period.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
19
Volume 18, 2023
3. Formation of data tables with information on
each trading candle: the opening price, closing
price, maximum value, minimum value, and
trading volume.
4. Data preparation in two stages (from the point
of view of the CRISP-DM methodology, these
stages are similar to Data Understanding and
Data Preparation):
• Data Engineering, which includes data
collection, understanding, as well as cleaning,
and initial data processing;
• Feature Engineering, which includes the
formation of descriptive features to data on
aspects of the behavior of the object whose
model is being built.
5. Training the model using the ZigZag indicator,
which shows how it was necessary to trade to
get maximum profit. Other technical analysis
indicators are also used at this stage. The
libraries of the R language are used to calculate
them.
6. Selection of features (feature selection), in other
words - variables based on which the model is
trained. The selection is carried out using
various tools. In addition, the dependence of
factors such as the correlation of features with
the target variable or the quality of data is
considered.
7. Creating new features based on existing ones
(feature engineering). This stage allows you to
improve the quality of the future model, while at
the same time getting a more complete
explanation of the data (for the interpreted
model). In our case, the first stage of building a
model in SAP Predictive Analytics was the
creation of new features using the built-in Data
Management solution.
8. Building a model in SAP Predictive Analytics
(automatic selection of variables should be
enabled).
9. Selection of variables with the smallest error in
the final equation. Calculation of the final
predictive strength of the model, robustness
(stability of the result to new data sets). These
values determine the quality and stability of the
model.
After passing the specified actions according to
the algorithm, the predictive analytics system
generates a forecast about further price movement in
a given period of time.
Our example demonstrates the capabilities of one
of the predictive analytics systems. Open sources,
[2], [9], [10] describe other systems of predictive
analytics of investment instruments that form a
forecast of price changes based on a particular set of
technical analysis indicators. The picture of the
results in the subject area under study is
complemented by several practical works, [11],
[12], [13], [14] describing the means of collecting
and analyzing investment market data and technical
analysis indicators using the libraries of the R and
Python languages.
Based on the generalization of information from
the above sources, it is concluded that price
movement forecasts and predictive analytics of
investment instruments are based on technical
analysis and analysis of historical data. It assumes
that external factors are embedded in the history of
price changes.
However, it should be noted that predictive
analytics systems do not consider the primary
source of changes. Also, they do not evaluate
external factors and events, the occurrence of which
is known, but which has not yet occurred. Of course,
after the fact, the price change in a certain period
will reflect the influence of an external factor. But
such a change is not tied to the cause (primary
source) of this factor. As a result, the predictive
analytics system can take a sharp price change for
an anomaly and throw it out of the training sample.
For example, such an external factor as a change
in the key refinancing rate dramatically affects the
price of an investment instrument both at a certain
point in time and in the long term. And this is only
one of the factors, their list and assessment of
significance are a separate topic for discussion, [15].
In practical terms, a predictive analytics system
should be able to predict a fall in the value of an
investment instrument in the short term due to the
influence of an external factor, but at the same time
predict a further upward trend in the long term.
Thus, the investor will be able to wait for the asset
price to fall, buy it at a more favorable price, and in
the long term receive a greater profit from dividends
or the sale of an investment instrument. In the
future, considering the predictive analytics system
of external factors will be able to minimize risks and
maximize profits for the investor.
The disadvantages of predictive analytics
systems include the fact that their forecast does not
evaluate price changes in a given period of time. As
a result, the analyst misses such important points as
the best price to buy an investment instrument and
the forecasted sale price of a trading asset. There is
a significant limitation - the system can help an
investor or trader to increase the number of
successful transactions, but not improve his trading
operations at a qualitative level.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
20
Volume 18, 2023
A typical methodology for implementing and
implementing a predictive analytics system consists
of three stages:
1. Preparatory work:
• Evaluation of available data;
• Definition of requirements, quality metrics, and
success criteria;
• Data transmission and dataset preparation;
• Forecast of economic effect.
2. Pilot project:
• Model development and training;
• Development of a system with an embedded
model;
• Model and system testing;
• Checking the success criteria, and evaluating the
effectiveness of the model.
3. Industrial operation of the system:
• Putting the system with an integrated model into
commercial operation;
• Validation of the service (A/B testing metrics),
including validation of the economic effect;
• Service support, including regular training.
The criterion of success is the achievement of the
system's performance targets. In the case of a
predictive analytics system, the key criteria for
success are the percentage accuracy of the forecast
and the percentage of correct forecasts.
At the same time, one should not lose sight of the
calculation that the predictive analytics system saves
time in searching for information, analyzing it, and
making decisions based on analysis.
3 Problem Solution
The initiator of the business requirements is the
customer of the predictive analytics system. As a
result of communication and requirements
collection, key business requirements for the
product were identified:
1. Implement a system of predictive analytics of
investment instruments to solve the problem of
inaccurate forecasts for investment instruments.
2. Reduce the staff involved in the analysis of
investment assets in the stock market.
Based on technical and fundamental analysis, a
financial analyst forms a forecast (investment
strategy), which describes the best entry and exit
points, and various scenarios for the development of
events.
We have identified the main bottlenecks in the
implementation of the business process. The first is
that the forecasts are subjective, despite the
availability of solutions and methods of working
with them. In open sources for one investment asset,
you can find dozens of different forecasts and
trading strategies from financial analysts, [16], [17].
Also, the bottleneck is the variability of price
change scenarios, which also brings a subjective
nature to analytics. And the last bottleneck is the
time error, because of which a financial analyst can
analyze a limited number of assets.
The implemented system using the developed
solution should produce a full technical analysis and
partially fundamental analysis of all assets, as well
as give an objective forecast of the price movement
of the selected asset. Thus, the user of the system
will receive an objective forecast with one scenario
and the probability of its execution. The number of
analyzed assets is unlimited. Such a system will
replace some of the employees involved in the
analysis of trading assets, as well as improve the
quality of forecasts, expressed in the number and
accuracy of forecasts.
Among the general requirements for a predictive
analytics system based on the developer solution,
we have identified the following. The system of
predictive analytics of investment instruments is
designed to predict the movement of the asset price
in a specified period of time. The user should be
able to view this information, as well as the ability
to configure the prediction time interval. It is
necessary to achieve accessibility and ease of
obtaining information on the movement of the price
of an investment asset.
All forecasts must be displayed correctly and
without errors on the on-screen forms of the
application. It is necessary to ensure the operation of
the system and update its data, including forecasts,
in real-time.
The system should allow the user to create
folders of selected trading instruments, and search
the system by keywords, and attributes. The
attributes that are searched for include: the name of
the investment instrument; the name of the
investment asset or its abbreviation.
The developed toolkit answers the following
questions:
1. What data needs to be collected.
2. How the data should be analyzed.
3. How the predictive model is built and trained.
4. What functionality is needed.
5. How the price movement is predicted.
To determine the required data set, it is necessary
to understand what relates to internal data and what
relates to external data. The internal data will
include all the data presented on the chart of the
movement of the trading instrument•
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
21
Volume 18, 2023
• Date;
• Time;
• Time period;
• Opening price;
• Closing price;
• Maximum price value;
• Minimum price value;
• Trading volume.
The above list of recommended data can be
requested from official trading exchanges via API
requests. The implementation of the API is
necessary in order for the predictive analytics
system to work with historical data (Table 1).
Table 1. Set of API methods of the predictive
analytics system.
With the selection of external data, the situation
is more complicated, since for different types of
trading instruments, there will be a different type of
external data. Let's give, for example, a list of data
for stock analytics:
• The exchange rate of the stock;
• Economic calendar based on the stock exchange
rate;
• Economic reports of the company;
• Availability of dividends on the stock;
• Supply and demand (glass of prices).
This dataset can also be extracted from various
sources that provide information in the form of
tables with data.
The difficulties that arise at this stage are the
need for text recognition to determine the type of
news in the economic calendar, as well as the need
to monitor market fluctuations in real-time. This
indicator changes together with the price change,
[18].
The next stage - exploratory data analysis (data
mining) involves the discovery of information in the
data. To do this, a sequence of operations should be
performed:
• Select the data (tables, records, and attributes);
• Clear the data, including performing their conversion
and preparation for modelling;
• Make derived data;
• Combine data;
• Bring the data into the desired format.
The essence of this stage is to sift through the
data, clearing them of anomalies. Additionally, in
the price series information, it is necessary to
analyse information on technical analysis indicators.
Of all the available indicators, one of the most
useful indicators for the model that will be trained in
the Zig Zag indicator (SAP Predictive Analytics). In
practical trading, it is not used to predict price
movements, but it shows how it was necessary to
trade to maximize profits. It is also used to generate
a price series. If the price was rising, then the
objective function in this period of time takes the
value 1, if it was falling, then 0. Thus, a mark-up for
a time series of prices is obtained (Figure 1).
Method
URL
Type
GET
/predict-analyz/initialize
Privat
GET
/predict-analyz/content
Privat
GET
/predict-analyz /select-disk
Privat
POST
/predict-analyz /upload
Privat
DELETE
/predict-analyz /delete
Privat
GET
/predict-analyz /preview
Privat
POST
/predict-analyz /create-
directory
Privat
POST
/predict-analyz /create-data
Privat
POST
/predict-analyz /update-data
Privat
GET
/predict-analyz /stream-data
Privat
GET
/predict-analyz /indicators
Public
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
22
Volume 18, 2023
Fig. 1: The "Zig Zag" indicator is shown as a red
stripe on the price chart.
The ZigZag indicator is designed to analyze
price movements with a given or greater amplitude
and is usually used by traders to visualize the trend -
it helps to highlight significant changes in the
quotation chart and ignore small fluctuations
(noises). This tool is used by many trading
platforms and is integrated as a function of software
from different vendors. In our model, the ZigZag
indicator is used to generate a price series to avoid
noise and, as a result, prevent retraining of the
model.
In addition to the price series, it is necessary to
calculate a set of values of technical analysis
indicators (Table 2).
Table 2. Set of values of technical analysis
indicators.
The choice of indicators of technical analysis
indicators is determined by the level of risk - the
riskier the operation is, the more indicators are
analysed. The result of the forecast is to buy, sell, or
wait for advice. Sets of indicators are used mainly
on thematic sites dedicated to trading instruments.
The accuracy of the forecast is fixed at the moment
for certain time periods. The time factor is of
decisive importance in this case, since in the case of
a strong price movement, the forecast may change
to the opposite.
As additional indicators (correcting own
calculations), it is recommended to supplement the
technical analysis with the following indicators:
• I43 (SMA24Trand) The logarithm of the
SMA24 ratio compared to the previous value;
• I44 (SMA60Trand) The logarithm of the
SMA24 ratio compared to the previous value;
• I45 (MOM24) Momentum 24 / Rate of Change;
• I46 (MOM60) Momentum 60 / Rate of Change;
• PC (PC 1-PC16) Compression of features I01-
I46 by the method of principal components in
16 values.
Input
Technical name
Technical Analysis Indicator
I01
EMA5Cross
The intersection of EMA 5 and opening prices
I02
EMA17Cross
The intersection of EMA17 and the opening price
I03
EMA5_17Cross
Intersection of EMA17 and EMA5
I04
VolumeROC1
Rate of Change / Momentum
I05
CCI12
Commodity Channel Index 12
I06
MFI14
Money Flow Index 14
I07
MOM
Momentum 3 / Rate of Change
I08
Lag1
Price movement at the current bar (1)
I09
Lag2
Price movement at the current bar (2)
I10
Lag3
Price movement at the current bar (3)
I11
Lag4
Price movement at the current bar (4)
I12
Lag5
Price movement at the current bar (5)
I13
fastK
Stochastic Fast %K
I14
fastD
Stochastic Fast %D
I15
slowD
Stochastic Slow %D
I16
stochWPR
William's %R
I17
RSI14
Relative Strength Index (open) 14
I18
williamsAD
Williams Accumulation / Distribution
I19
WPR
William's %R 14
I20
AO
(Awesome Oscillator, AO) SMA5 — SMA34
I21
AC
AO smoothed 5-period average AO —SMA(AO, 5)
I22
MACD
EMA12 – EMA26
I23
MACD_SMA9
MACD smoothed by a 9-period moving average MACD-
SMA(MACD, 9)
I24
DIp
The positive Direction Index
I25
DIn
The negative Direction Index.
I26
DX
The Direction Index
I27
ADX
The Average Direction Index (trend strength)
I28
Ar
aroon(HL, n) – 1 out (oscillator)
I29
chv16
Chaikin Volatility – chaikinVolatility (HLC,n) -1 out
I30
cmo16
Chande Momentum Oscillator – CMO(Med, n) -1 out
I31
macd12_26
MACD Oscillator 12, 26, 9
I32
Osma
Moving Average of Oscillator
I33
rsi16
Relative Strength Index med 16
I34
fastK14_3_3
Stochastic Oscillator 14 3 3 fastK
I35
fastD14_3_3
Stochastic Oscillator 14 3 3 fastD
I36
slowD14_3_3
Stochastic Oscillator 14 3 3 slowD
I37
smi13_2SMI
Stochastic Momentum Index SMI 13 2
I38
smi13_2signal
Stochastic Momentum Index signal 13 2
I39
vol16
Volatility 16
I40
SMA24Cross
Logarithm of the ratio of the opening price and SMA24
I41
SMA60Cross
Logarithm of the ratio of the opening price and SMA60
I42
SMA24_60Cross
The logarithm of the SMA24 and SMA60 relationship
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
23
Volume 18, 2023
It is also necessary to collect and compare the values of
economic indicators used in fundamental analysis
(Table 3).
Table 3. Values of economic indicators used in
fundamental analysis.
The use of economic indicators for fundamental
analysis has one single purpose — to identify
securities that are currently very cheap or expensive,
both relative to competitor securities and relative to
their estimated fair value. Fundamental analysis
based on the comparison of the values of economic
indicators helps to adjust the strategy of selecting
technical indicators when building a predictive
analytics model. Fundamental analysis is a tool for
"long" investments, as it is based on the following
provisions:
• The current price of the investment object
cannot objectively reflect the real value of the
company;
• In the long term, the stock market tends to
bring the market value closer to the true value.
To build a future predictive model, it is required
to select features. In the prepared dataset, there are
indicators of indicators that affect the target variable
at the current time. However, you can get additional
information if you determine the impact of these
indicators for a certain period up to the current
moment. According to our toolkit, time intervals
were selected: 1 hour and 1 day before the current
moment in time and new variables were created
considering this "time lag".
Even more, information may be the degree of
change in indicators from the moment in the past to
the current moment. The natural logarithm of the
quotient of the current indicators and indicators with
a time lag of 1 hour and 2 days can be chosen as a
method. Thus, it will be possible to obtain the
degree of change of the indicator from the moment
in the past (it increased or decreased), and if so, how
much.
Accordingly, these steps will allow us to
establish not only the relationship between the
current values of the indicators and the target
variable but also to consider the state of these
indicators in the past, as well as the degree of their
change.
As a basic model for the implementation of the
system, it is proposed to use a linear regression
model or a polynomial regression model, which will
reveal the overfitting of the model.
As an implementation solution according to the
toolkit, it is proposed to use Python programming
languages using libraries: model-catwalk, sklearn,
and the like. As an alternative, it is possible to use
the R programming language with packages:
AppliedPredictiveModeling, LiblineaR, astsa, and
data.table, timeSeries, eclust.
The training of the model should be carried out
on the prepared data. At the same time, the
percentage of correct predictions during testing and
training is accepted at least 70%.
The practice of using a trained predictive model
is not presented in this study, since these results are
a trade secret. However, the algorithm itself for data
analysis and model construction is an open-to-use
solution, that can be repeated and scaled.
4 Conclusion
This study describes the process of developing a
solution that considers the main shortcomings of
approaches to implementing IT solutions in the field
of predictive analytics investment processes. The
toolkit is based on the main components and
principles of the standard methodology,
supplemented and reworked for the object of
research.
The developed solution does not consider many
issues related to system testing and metrics
evaluation, since the main task was to develop a
solution for the implementation of the system. This
solution will make it possible to make a qualitative
selection of data for the analysis and training of the
predictive analytics model of investment
instruments.
The proposed solution will: improve the quality
of forecasts (increase the accuracy of the forecast);
minimize the costs of an investor or consultant
company for the analysis of investment instruments
(reduction of analysis time and optimization of
labour resources); increase the objectivity of
forecasts (rational choice of qualitative and
quantitative indicators); simplify the process of
analysing investment instruments for users of the
information system (interpretation of results
analytical evaluation).
The theoretical significance of this study lies in
the fact that the fulfilment of the tasks set made it
possible to generalize and structure information
about the technologies used in the implementation
of information systems using predictive analytics.
Input
Technical
name
The name of the indicator
F01
T
Total revenue for the period
F02
P
Net profit for the period
F03
EBITDA
Profit before deduction of interest, taxes, depreciation and
amortization expenses
F04
act
Assets
F05
obl
Obligations
F06
Y
Market capitalization
F07
P/E
Price/earnings per share ratio
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
24
Volume 18, 2023
The conclusions made may be useful to future
researchers as the initial data of independent
scientific research.
If the research process itself is more focused on
achieving theoretical significance and allows you to
demonstrate the depth and scale of the work carried
out, then the developer solution for implementing
the predictive analytics information system is of
applied importance. The practical significance of the
results obtained lies in the fact that they can be used
by financial organizations engaged in investment
activities, such as banks, investment funds, private
investment companies, as well as individuals
engaged in investing funds through brokers.
References:
[1] Letourneau-Guillon L, Camirand D, Guilbert F
and Forghani R, Artificial Intelligence
Applications for Workflow, Process
Optimization and Predictive Analytics,
Neuroimaging Clin. N. Am. 30, 2020, e1–15
[2] Sezer O B, Gudelek M U and Ozbayoglu A M,
Financial time series forecasting with deep
learning: A systematic literature review: 2005–
2019, Appl. Soft Comput. 90, 2020, 106181
[3] Gupta S, Drave V A, Dwivedi Y K, Baabdullah
A M and Ismagilova E, Achieving superior
organizational performance via big data
predictive analytics: A dynamic capability
view, Ind. Mark. Manag. 90, 2020, 581–92
[4] Sadeghi Eshkevari S, Cronin L, Sadeghi
Eshkevari S and Pakzad S N, Input estimation
of nonlinear systems using probabilistic neural
network, Mech. Syst. Signal Process. 166,
2022, 108368
[5] Oo M C M and Thein T, An efficient predictive
analytics system for high dimensional big data,
J. King Saud Univ. Comput. Inf. Sci., 2019
[6] Markitanov D V, Integration of predictive
analytics system with industrial network,
Novosibirsk State Technical University, 2019
[7] Kashyap S, Corey K M, Kansal A and Sendak
M, Machine learning for predictive analytics,
Mach. Learn. Cardiovasc. Med., 2021, 45–69
[8] Nisbet R, Miner G and Yale K, The Data
Mining and Predictive Analytic Process,
Handb. Stat. Anal. Data Min. Appl., 2018, 39–
54
[9] Dinis D, Teixeira P and Barbosa-Póvoa A,
ForeSim-BI: A predictive analytics decision
support tool for capacity planning, Decis.
Support Syst. 131, 113266, 2020
[10] Gulay E and Duru O, Hybrid modeling in the
predictive analytics of energy systems and
prices, Appl. Energy 268, 114985, 2020
[11] Weng B, Lu L, Wang X, Megahed F M and
Martinez W, Predicting short-term stock prices
using ensemble methods and online data
sources, Expert Syst. Appl. 112, 2018, 258–73
[12] Ozbayoglu A M, Gudelek M U and Sezer O B,
Deep learning for financial applications: A
survey, Appl. Soft Comput. 93, 2020, 106384
[13] Zhou M, Financial auditing big data platform
based on FPGA and convolutional neural
network, Microprocess. Microsyst., 2020,
103461
[14] de Oliveira Carosia A E, Coelho G P and da
Silva A E A, Investment strategies applied to
the Brazilian stock market: A methodology
based on Sentiment Analysis with deep
learning, Expert Syst. Appl. 184, 2021, 115470
[15] Prince J T, A paradigm for assessing the scope
and performance of predictive analytics, Inf.
Econ. Policy 47, 2019, 7–13
[16] Wang C, Zhang X, Wang M, Lim M K and
Ghadimi P, Predictive analytics of the copper
spot price by utilizing complex network and
artificial neural network techniques, Resour.
Policy 63, 2019, 101414
[17] Gu X, Mamon R, Duprey T and Xiong H,
Online estimation for a predictive analytics
platform with a financial-stability-analysis
application, Eur. J. Control 57, 2021, 205–21
[18] Loshin D, Knowledge Discovery and Data
Mining for Predictive Analytics, Bus. Intell.,
2013, 271–86.
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
The authors equally contributed in the present
research, at all stages from the formulation of the
problem to the final findings and solution.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
No funding was received for conducting this study.
Conflict of Interest
The authors have no conflicts of interest to declare
that are relevant to the content of this article.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.2
Natalia Mamedova, Olga Staroverova,
Alexey Epifanov, Huaming Zhang, Arkadiy Urintsov
E-ISSN: 2224-2856
25
Volume 18, 2023