Data Technology Apply in Business Decision Making:

How to Use Data Information to Make Business Decisions in the

Digital Age

Xingxian Li

63-1-1001 Xinxian Jiayuan, Huairou District, Beijing, 101400 Beijing, China

Keywords: Business Decisions, Data-Driven, Data Capature, Data Cleansing and Standardization, RPA Technology, The

Normalized, Z – Score, Fitting, Data Conversion.

Abstract: Data are an important asset of an enterprise, changing the way information is connected and reshaping the

future of the enterprise. To make business decisions quickly and without mistakes, business managers must

rely on data and information. The construction of enterprise data information will follow the DIKW model,

the seven-step process rule, the four-layer technical framework and the underlying powerful algorithm

foundation. Data capature, data cleaning and standardization, and data modeling are the basis for making

business decisions based on data information. This academic paper will talk about the basic methods of data

standardization and data modeling. It comes from working practice.

1 INTRODUCTION

Business decision is a process in which enterprises or

organizations make decisions on future actions after

analyzing, calculating and judging the factors

affecting the realization of goals based on objective

possibilities and certain tools, skills and methods with

the help of certain information and experience. For

business managers, management is decision-making.

"Decision-making occupies the core position in

management activities and runs through the entire

process of management activities." (Li, 2020)

Efficient and high-quality decision-making drives

the enterprise to continuously provide high-quality

products (or services), making the enterprise stand out

and win in the fierce market competition. In today's

digital era, the speed and amplitude of change are far

beyond the past, which puts forward higher

requirements on the decision-making ability of

enterprises, and the decision-making must be both

good and fast. However, in the real business world, the

decisions made by enterprise managers fail to reach

the expected goals due to the lack of systematic

methods. "According to Microsoft, more than 74

percent of business decisions are behind schedule or

fail......" jean-Paul Sartre wrote in his book The

Difficulty of Making Decisions. The actual situation

is so bad, what kind of decision-making mechanism

can provide enterprise managers with ways and means

to get out of the decision-making dilemma? Based on

years of enterprise management experience and

systematic learning summary, the article author has

been studying the management advantages of

advanced enterprises in recent years, and comes to the

conclusion that only by relying on digital technology

can we make high-quality decisions and avoid

decision-making mistakes. The digital technology

mentioned here is not only the summary of the past

data information and experience, but also the

simulation and prediction of the future trend. For

example, Amazon's personalized recommendation

based on big data derives more than one third of its

revenue from recommendation functions. McKinsey

defines "Industry 4.0" as the digitization of

manufacturing, with sensors embedded in almost all

components and equipment, and the widespread

introduction of cyber-physical systems that analyze all

available data. Relying on digital technology, senior

managers, middle managers and first line managers

respectively focus on strategic decisions, management

decisions and daily rountine operation decisions, and

they play their own role to ensure the continuous and

efficient operation of the enterprise and its survival in

the ever-changing business world.

Li, X.

Data Technology Apply in Business Decision Making: How to Use Data Information to Make Business Decisions in the Digital Age.

DOI: 10.5220/0012028900003620

In Proceedings of the 4th International Conference on Economic Management and Model Engineering (ICEMME 2022), pages 247-256

ISBN: 978-989-758-636-1

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

247

2 DATA-DRIVEN BUSINESS

DECISIONS ARE AN

INEVITABLE CHOICE FOR

ENTERPRISES

2.1 Definition of Business Decisions

Talking about business decision, let's take a look at its

definition first: Business decision is made by the

relevant organization of the enterprise to enhance the

strength of the enterprise, improve the profitability of

the production and operation of the decision. Business

decision-making mechanism is a mechanism by which

enterprise managers make decisions on production,

management and other business activities under the

condition of having sufficient legal person property

rights. The business decision mechanism is in the

main position in the operation mechanism. It is not

only the basis of designing other mechanisms, but also

runs through the operation of other mechanisms. A

sound decision-making mechanism is a necessary

condition for effective decision-making.

2.2 The Main Points of Decision Theory

Since the second World War, many operational

researchers, statisticians, computer scientists and

behavioral scientists have been trying to find a

scientific way of making decisions in the field of

management in order to make clear and rational

choices on complex multi-scheme problems. With the

study of this aspect, decision theory has been

developed rapidly. Decision Theory School is an

emerging management School based on statistics and

behavioral science and using computer technology

and research methods. The main representatives of

traditional theories are Herbert Simon, A. Simon and

James G. March. The core theory is the decision

theory proposed by Herbert Simon, and the main

viewpoints are as follows:

Figure 1: Three stages of business decisions rely on.

• Management is decision-making: Simon et al.

believe that the whole process of management

activities is decision-making process. Determining

objectives, making plans and selecting plans are

business plans and planning decisions; Mechanism

design, production unit organization and authority

allocation are organizational decisions; Inspection of

plan execution, wIP control and selection of control

means are control decisions. Decision-making runs

through the whole management process, so

management is decision-making.

• Decision making is divided into procedural

decision making and non-procedural decision making:

procedural decision making refers to the decision

made in accordance with established procedures;

Special treatment is required when the problem is

widespread, new, unstructured, or so important and

complex that there is no routine procedure to follow.

Decisions on such questions are called non-procedural

decisions.

• Satisfactory code of conduct: Simon thinks,

because the organization under the changing external

environment influence, to collect all the data which are

ICEMME 2022 - The International Conference on Economic Management and Model Engineering

248

difficult, and some action plans are more difficult to

list the sites, and the person's knowledge and ability is

limited, so when making decision, it is difficult to

obtain the best solution, in practice, even able to find

out the best solution, in economic terms have to

consider, People also tend not to pursue it, but to make

decisions based on satisfying principles.

With the development of digital technology,

modern decision-making theory has gradually come

into being, represented by Venkat Venkatraman. In his

book The Digital Matrix: New Rules for Business

Transformation through Technology, he provided

three decision-making methods conducive to success:

first, how to avoid getting lost in the dynamic

ecosystem: carefully build and participate in various

ecosystems; The second is how to work with different

companies to build new capabilities and create new

business value: to connect with competitors and

potential Allies; Finally, how to design the

organizational structure to reflect the new and

powerful model of human-computer interaction: using

powerful machines to amplify the enterprise's

potential.

2.3 The Three Stages of Development of

Business Decisions

With the application of new advanced informatization

and digital technologies such as artificial intelligence,

cloud computing, big data, blockchain, Internet of

Things, and the Internet, the degree of digitization of

society continues to increase, and data have become

an important element in building a modern society.

From 2020, data have become the fifth largest factor

of production after land, labor, capital and technology.

It is an important asset of enterprises and, of course,

an important asset of individuals, organizations and

even countries. "In the digital age, companies need to

have a new understanding of data, because data has

become the new core asset." (Ram, 2020) Business

decisions based on data information will not affect the

quality of decision-making due to the subjective

factors of managers, and avoid decision-making

mistakes or major decision-making mistakes.

"Information and value form the 'foundation' of

decision-making: what we can do, what we know, and

what we want." (Carl, 2017) The evolution process of

the basis for supporting decision-making is shown in

Figure 1: The three stages of the business decision-

making model are the empirical decision-making

model, the electronic information decision-making

model and the data-driven decision-making model.

Since 2013, enterprises have entered the data-driven

decision-making model. The data-driven decision-

making model reflects that the entire business chain of

the enterprise's R&D, planning, organization,

production, coordination, sales, service and

innovation uses digital decision-making, and supports

the strategic decision-making and planning of the

entire enterprise, enabling the enterprise to achieve

overall Decision intelligence, and ultimately lead the

transformation of enterprises and even the industry

through data-driven. The Fraunhofer Institute in

Germany put forward the concept of Industry 4.0. The

institute believes that the logical starting point of

Industry 4.0 is to adapt to the rapid changes in the

competitive environment. How does an enterprise

adapt to the rapid changes in the market + users +

products + technology, it can be seen that the

traditional low-frequency decision-making

mechanism cannot adapt to the high-frequency

decision-making needs in emergencies, and data-

driven fast and high-quality decision-making is an

inevitable choice for modern enterprises.

3 THE FORMATION PROCESS

OF BUSINESS DECISION

DRIVEN BY DATA

INFORMATION FROM DIKW

MODEL

Just like the accumulation of knowledge, the

formation of data-driven decision mechanism is

essentially a process of enterprise capacity building.

Enterprise-related data are collected, processed,

identified, processed and presented, and finally

become the knowledge and wisdom to guide

enterprise operation and management. This process is

presented by DIKW model as shown in Figure 2,

which is to understand this process from the cognitive

level.

3.1 The Relationship Between DIKW

Model and Enterprise to Make

Intelligent Business Decision

DIKW shows the universal process of evolution from

data, information, knowledge to wisdom. This model

will also be followed by intelligent decision-making

based on data information from the enterprise.

• D-Data: Data can be numbers, words, images,

symbols, etc. It comes directly from facts and can be

obtained through original observation or

measurement.

Data Technology Apply in Business Decision Making: How to Use Data Information to Make Business Decisions in the Digital Age

249

Figure 2: DIKW model and business Intelligence decision process.

• I-Information: by organizing and processing

data in a certain way and analyzing the relationship

between data, data become meaningful, which are

Information. Information can answer simple questions

such as: Who? What? Where to? What time? So

information can also be thought of as data that is

understood.

• K-knowledge: Knowledge is useful

information filtered, refined and processed from

relevant information. It is a collection of information

that makes information useful. It is a process of

judging and confirming information, which combines

experience, context, interpretation and reflection.

Knowledge establishes meaningful connections

between data and information, and between

information and the application of information in

action. It embodies the essence, principles and

experience of information. Knowledge answered,

"Well?" Problems to help enterprise modeling and

simulation.

• W-Wisdom: Wisdom is an extrapolated,

nondeterministic, nonjudgmental process. Unlike

previous stages, wisdom focuses on the future, trying

to understand what was not understood or done in the

past. Wisdom can be summarized as the ability to

make sound judgments and decisions, including the

best use of knowledge. Wisdom answers the question

"Why?"

3.2 Smart Business Decisions Take

Time to Build

It takes time, maybe a year or even a few of years,

from data capature to the extraction of information

then to intelligent decisions of enterprises. First of all,

data capature is complicated, because data sources are

online, offline, inside and outside the company,

including historical data and model-based forecast

data. Secondly, after the data are obtained, how to

organize the data to make it information, involves data

cleaning, sorting, association and other technical

problems; Thirdly, the information obtained from data

capatureis massive and needs to be processed,

extracted and abstracted. This process involves the use

of various analytical methods, and it is a process that

is gradually deepened with business insight. Only

when enterprise managers' cognitive level reaches a

certain level, and with the assistance of IT technology,

can they be data-driven. Digital decision making

focuses on the automation and optimization of specific

business decisions. (Venkat, 2018)

ICEMME 2022 - The International Conference on Economic Management and Model Engineering

250

4 HOW IS THE DATA-DRIVEN

DECISION PROCESS

IMPLEMENTED FROM THE

PERSPECTIVE OF PROCESS

CONSTRUCTION

The process of implementing data-driven decision-

making is divided into 7 levels, as shown in Figure 3

below:

Figure 3: The 7 stages of data-driven enterprise business

decision making.

1) Basic IT system: The first level is the "basic IT

system", which is a data-driven foundation, and its

function is to complete data capature. It mainly refers

to the software system and its supporting hardware

equipment used by the enterprise in the actual

operation process, such as various business systems

of the enterprise, financial management software,

CRM system; hardware equipment such as sensors,

detectors, etc., these systems complete "data

capature" task. "It can be said that the best, latest and

most flexible ideas come from the end." (Tencent,

2020) The raw data obtained after data capature are

non-standard and unusable. They must be processed

before they are meaningful, thus entering the next

stage of "data cleaning and standardization".

2) Data cleaning and standardization: At the "data

cleaning and standardization" level, what we are

trying to achieve is to break down data barriers so that

data can flow normally within the enterprise. (Wang,

2020) The work of this stage covers: a) data cleaning;

b) Data integration; c) Data distribution and

transformation; d) Data reduction and other pre-

processing work.

3) Data reporting and Visualization: The question

in this step is: How do you make the data visible? The

simplest and most straightforward method is "data

reporting." It is to construct various forms according

to daily business usage and fill in a large amount of

data in the forms. Some enterprises make reports

manually, some enterprises use report engine to make

reports, and some enterprises enter the stage of data

analysis and visualization. Through BI and other

analysis tools, they delegate the right of data analysis

to the user end, helping the business to quickly get

data and quickly make reports, and even do some

analysis independently. From "basic IT systems" to

"data reporting and visualization," the first three

levels are, in some ways, the foundation for data

analysis and application. For an enterprise, complete

the three levels, some companies are done manually,

some companies are localized deployment of IT

systems, some companies is done by the cloud IT

systems, only the three levels of ability, can be said

that the enterprise has the use of data to guide the

operation, the decision-making, management and so

on the basis of data applications.

4) Product and operation analysis: the first target

is the monitoring of daily operation; Second, when

the daily analysis has become a routine part of the job,

the enterprise products and business people will find

simple daily analysis cannot solve the problem of

complex management and strategic decision, unable

to bring a surprise to customers, which requires the

user, product, channel, market, demand, and so on

aspects of deep analysis and research. In this process,

many business-specific analysis topics and data

models have emerged to help enterprises better

understand the market, and capture customers and

potential business opportunities. The most

representative example of this is "user portraits".

5) Lean operation: At the level of "lean

operation", all analyses are no longer isolated from

each other, but more based on an actual business

scenario to realize the overall management of all

processes in this scenario. If multiple applications or

systems can be built in each field of the enterprise,

then these aggregations can basically support the

main enterprise operation and management.

6) Data product: Data mining is an evolutionary

product generated by enterprise data, and it is one of

the many ways for enterprises to realize the value of

internal data. Data products in the physical industry

are often due to the fact that the internal data

capabilities of the enterprise have grown to a certain

stage, and some internal data and analysis methods of

the enterprise have already met the conditions for

independent realization, so they are taken out by the

enterprise as a type of product and provided to the

market. Data products are formed, and the data

products of entity enterprises serve more within the

organization.

7) Data Strategy: Companies use data

strategically to accelerate decision-making through

Data Technology Apply in Business Decision Making: How to Use Data Information to Make Business Decisions in the Digital Age

251

business insights and ultimately achieve their

strategic goals. After long-term scientific

governance, data have become a strategic resource for

enterprises, and data information are used to gain

competitive advantages and achieve business goals.

5 SEE HOW DATA-DRIVEN

DECISION-MAKING GOALS

ARE ACHIEVED WITH THE

TECHNICAL FRAMEWORK

OF IT

In order to achieve data-driven decision-making goal,

the technical architecture of the enterprise system is

divided into four levels: data capture → data

standardization → mining data value → making

intelligent decisions, as shown in Figure 4:

Figure 4: A 4-tier technical framework for data-driven business decisions

Data capture: The data API management platform

uses RPA technology for data aggregation, integrates

the data scattered in various information islands, and

displays the data source and data call records at the

same time to ensure the stability of data calls.

Real-time data processing (cleaning,

standardization, etc.), data quality management: CEP

stream computing software, powerful stream data

computing capability, can handle complex events;

massive data throughput, millisecond-level response.

Data mining: Machine learning platform, with

powerful integration capabilities of data and tools, can

be easily expanded; general technology can preset

more than 2,000 modules of Hull's advanced

algorithms, auxiliary modeling, etc.

Intelligent business decision: decision engine

platform; timely early warning and rapid response

operation monitoring system; data visualization

system. FICO, IBM, Experian and other technologies

achieve easy-to-use, agile, and intelligent effects.

6 EXAMPLE: DATA CLEANING,

STANDARDIZATION

METHODS

The previous part succinctly shows the IT technical

architecture that data support business decision-

making. Data collection can rely on various software

systems, hardware equipment established by the

ICEMME 2022 - The International Conference on Economic Management and Model Engineering

252

enterprise in the past, or even collect data manually,

and complete the aggregation through RPA robots.

The aggregated data are waiting for cleaning and

standardization. This process cannot be done

manually and must rely on technology. “Data by

themselves do not express any meaning, they are only

when data are combined with logic that we can

discover and express insights.” (Zhou, 2021)

Therefore, this chapter will introduce data cleaning

and standardization methods, which are part of big

data processing and come from work practice. The

methods are as follows:

6.1 Data Normalization

1) Data normalization: Data normalization is to map

the dimensional features of the data into a specified

range, [0, 1] or [-1, 1], and a compression dimension.

Normalization types can be divided into:

(1)

• Max-min normalization, the equation is as (1),

where X

old

is the original data set with m samples and

n features. min(X

old

) and max(X

old

) represent the

extreme values for each feature of the original data set.

• Mean normalization, the equation for mean

normalization is as (2), where mean(X

old

) represents

the mean value of each feature of the original data set.

(2)

• Non-linear normalization: Take the logarithm

of the original data. Non-linear normalization does not

scale the dimensional features of the original data to a

certain range, but reduces the scale (dimension) of the

dimensional features. Usually in some data

processing, the logarithm of the original data is often

taken before further processing. The reason for this is

that the logarithmic function is a monotonically

increasing function in its definition domain. After

taking the logarithm, it will not change the nature and

correlation of the data, and it can also compress the

scale (dimension) of the feature.

2) Additional notes on data normalization methods:

Features: Normalization will change the data

distribution of the original data, and the original

information is not preserved. The purpose of scaling

different features is to make the influence weights of

each feature dimension on the objective function to be

the same. At the same time, due to the different

degrees of scaling and transformation for different

features, the projected contour lines of those flat

distribution objective functions tend to be circular,

which also changes the distribution type of the original

data.

Functions: a) Speed up the training: such as the

convergence speed of the objective function in the

iterative algorithm; b) Balance the weights of the

features in each dimension to avoid the interference of

the features with too large or too small a numerical

scale on the model

Disadvantage: After normalizing the data,

although the weight of each dimension is balanced, it

also changes the data distribution of the original data,

that is, destroys the data structure.

6.2 z-score

Scale the data to a data distribution centered at 0 and

a standard deviation of 1 (Note: a data distribution

with a mean of 0 and a standard deviation of 1 is not

necessarily a normal distribution, it may also be a t

distribution or other distribution), in addition, z-score

retains the original data information and does not

change the original data distribution type. The purpose

of the z-score is also to make different features of the

raw data comparable. The z-score equation is shown

in (3)

(3)

In the equation, µ is the vector of the mean of each

column feature of the original data set, µ=mean

(Xold), σ is the vector of the standard deviation of

each feature of the original data set.

Comparison of data normalization and z-score:

The same aspect of normalization and z-score:

both perform linear transformation on the original

data, that is, both translate the sample points and then

shorten the distance, so that the different features of

the original data are comparable

The difference between normalization and z-score:

a) the impact of normalization on the objective

function is reflected in the value, while the impact of

z-score on the objective function is reflected in the

geometric distribution of the data; b) normalization

changes the amount of data level and also change the

distribution of the data, Z-score only changes the

magnitude of the data but does not change the

distribution type of the data; c) z-score normalizes the

data, does not change the contour projection of the

objective function, and will Continue to maintain the

Data Technology Apply in Business Decision Making: How to Use Data Information to Make Business Decisions in the Digital Age

253

flatness of the original objective function, and

normalize the data to make the contour projection of

the objective function appear circular; d) In the

gradient descent algorithm, normalizing the data helps

to speed up the convergence of the algorithm.

Data normalization and z-score usage scenarios:

• Using gradient descent parameter estimation

model: using normalized data can improve the

convergence speed of the algorithm

• PCA dimensionality reduction algorithm

needs to be decentralized, so Z-Score processing can

be used

• For specific requirements on value range, data

should be normalized, such as image processing,

where pixel intensity must be normalized to fit a

certain range (RGB color range 0 to 255).

• Probabilistic models are insensitive to feature

dimensional differences and can be standardized

without using measurement indicators (such as

decision trees).

In general, z-Score processing is used when it is

uncertain which data processing method to use,

because it does not change the data distribution type,

that is, does not break the data structure.

6.4 Centralization/Zero Meanization

After centralizing the data, the mean value of the data

is a 0 vector, which is to translate the original data to

the vicinity of the origin. The centralized processing

of data is a process of translation one by one, and does

not change the type of data distribution. Suitable for

PCA dimensionality reduction algorithm. The

centralized preprocessing expression is as (4)

(4)

The function is to facilitate the calculation of the

covariance matrix, remove the influence of the

intercept term (bias term), and increase the

orthogonality of the basis vector.

Figure5Fitting of Linear Regression.

6.3 Regularization

Regularize the data to scale a certain norm (L1 norm,

L2 norm) of each sample to 1, that is, calculate its p-

norm for each sample, and then for each element in the

sample Divide by this norm such that the p-norm of

each sample of the processed data is equal to 1. The

equation is as (5)

(5)

Regularization processing data is mainly used in

text classification and clustering, and has a great effect

on the need to calculate the similarity between

samples, such as (6) calculating the cosine similarity

of sample X

and sample X

(6)

7 EXAMPLES: DATA

MODELING, MACHINE

LEARNING METHODS

Machine learning is for modeling, so as to predict the

future under the built model and guide decision-

ICEMME 2022 - The International Conference on Economic Management and Model Engineering

254

making. Here are some basic methods of machine

learning:

7.1 Polynomial Features

Polynomial feature is a way to increase the dimension

of data. In linear regression, when using simple X

and

features to fit the curve, it cannot be completed -

underfitting, but we can create new features such as X

to fit the data, a better model may be obtained, so we

sometimes do a polynomial process on the features,

that is, change the features X

, X

into X

, X

, as

shown in the following figure

7.2 Data Conversion

According to the central limit theorem in probability

theory, when the sample size is infinite, the limit of

many distributions is the normal distribution. Many

random variables in reality are formed by the

combined influence of a large number of independent

random factors, and each of these factors plays a small

role in the overall impact. Such random variables tend

to approximate a normal distribution. (Objective

background to the Central Limit Theorem).

From the point of view of entropy (used to measure

the degree of confusion of information), the entropy of

the normal distribution is the largest among all other

distributions when the mean and variance of the data

are known (the original data distribution type is

unknown). According to the entropy standard,

"maximum entropy" is approximately equivalent to

"the closest uniform distribution under the same

constraints", that is, it is more practical. It can be

understood in this way that "entropy maximization" is

to make the ideal closer to reality, let the special

approach the general, and thus make the model more

general. Note that the entropy of the normal

distribution is actually determined by the variance,

and the "maximum entropy of normal variables" is a

conclusion in the context of a fixed variance. Different

variances obviously lead to different normal

distributions, and a normal distribution with higher

entropy has more variance - and is also closer to

"uniform" on the real axis.

Many machine learning models use normal

distributions, such as linear regression machine

learning models that require data features to be

normally distributed. If the data features are not

normally distributed, sometimes it is necessary to find

a mathematical transformation to transform the

features according to the normal distribution. Methods

as below:

1) Logarithmic transformation: For data

distributions that are highly skewed (eg, Skewness is

more than 3 times its standard error), we can take

logarithmic processing. Among them, it can be

divided into natural logarithm and logarithm with

base 10. Among them, logarithm with base 10 has the

strongest correction force, but sometimes it is

overcorrected and converts positive skewness into

negative skewness. The equation is as (7)

X = log(Xold) (7)

2) Square root transformation: The square root

transformation normalizes the samples that obey the

Poisson distribution or the samples with mild

skewness, or when the variance of each sample is

positively correlated with the mean, the square root

transformation can be used to make the variance

homogeneous sex. The expression is as (8)

Ξ=



𝑋



(8)

3) Reciprocal transformation: It is often used for

data with large fluctuations at both ends of the

distribution. The reciprocal transformation can

reduce the influence of extreme values. The

expression is as (9)

Ξ=1/Ξ

ολδ

(9)

4) Square root inverse rotation transformation:

commonly used for data subject to binomial

distribution or percentage. It is generally believed that

the equal overall rate is small (such as <30%) or large

(such as >70%), and the deviation from normality is

more obvious. Through the inverse transformation of

the square root of the sample rate, the data can be

close to the normal distribution, and the variance can

be achieved. homogeneity requirements. The

expression is as (10)

Ξ=ασιν(



𝑋



) (10)

5) BOX-COX transformation: usually used when

the continuous response variable does not meet the

normal distribution. In some cases (P value of

characteristic distribution < 0.003) the above methods

(square transformation, etc.) are difficult to achieve

normalization, so Box-Cox transformation can be

considered, but when P value > 0.003, using both

methods Yes, the ordinary square transformation is

preferred. (About the technical example, write it here

first, and we will discuss it together when we have the

opportunity)

Data Technology Apply in Business Decision Making: How to Use Data Information to Make Business Decisions in the Digital Age

255

8 CONCLUSIONS

Standardized data information is the basis for business

decisions. Data information is first and foremost a

perceptron of the environment. With the help of

technologies such as artificial intelligence, they work

side by side with corporate decision makers to make

up for human shortcomings and help the entire

organization run efficiently.

ACKNOWLEDGMENT

My sincere thanks to:

1. The knowledge taught by the tutors of the

Russian Friendship University makes my knowledge

a system;

2. The superior managers, colleagues, and

subordinates I meet in my work are the support,

guidance, and tolerance you have given me in practice.

It is not an exaggeration to use the idiom "a thousand

tempering";

3. Guidance from the reviewers of "2022 4th

International Conference on Economic Management

and Model Engineering (ICEMME2022)";

4. Strong support from my family.

REFERENCES

Audio books - In-depth Interpretation RPA, Reader Wang

Yan, Produced by Huazang, September 2020.

"DECISION QUALITY: VALUE CREATION FROM

BETTER BUSINESS DECISIONS" authors: Carl

Speitzler, Hannah Winter, Jennifer Meyer. Hunan

Literature and Art Publishing House, 2017

"McKinsey structured Strategic Thinking", author: Zhou

Guoyuan, Posts and Telecommunications Press,

January 2021.

"Modern Enterprise Management", authors: Li Dongjin,

Qin Yong, Chen Shuang, Posts and

Telecommunications Press, September 2020.

"Super Connection", Author: Tencent Smart Retail, Citic

Press, August 2020

"The Amazon Management System" authors (USA) Ram

Charan, (China) Yimei Yang. Machinery Industry

Press, 2020

"The Digital Matrix: New Rules for Business

Transformation through Technology", author: Venkat

Venkatraman, Guangdong People's Publishing House,

2018

ICEMME 2022 - The International Conference on Economic Management and Model Engineering

256