sustainability
Article
A Regional Industrial Economic Forecasting Model Based on a
Deep Convolutional Neural Network and Big Data
Shouheng Tuo 1,2,3, * , Tianrui Chen 1 , Hong He 4, *, Zengyu Feng 1 , Yanling Zhu 1 , Fan Liu 1 and Chao Li 1






Citation: Tuo, S.; Chen, T.; He, H.;
Feng, Z.; Zhu, Y.; Liu, F.; Li, C. A
Regional Industrial Economic
Forecasting Model Based on a Deep
Convolutional Neural Network and
Big Data. Sustainability 2021, 13,
12789. https://doi.org/10.3390/
su132212789
Academic Editor: Carlos
Rodríguez Monroy
Received: 26 October 2021
Accepted: 17 November 2021
Published: 19 November 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Sustainability 2021, 13, 12789. https://doi.org/10.3390/su132212789
1 School of Computer Science and Technology, Xi’an University of Posts and Telecommunications,
Xi’an 710121, China; ctr1498@126.com (T.C.); fengzengyu@stu.xupt.edu.cn (Z.F.);
zhuyanling11@126.com (Y.Z.); l_nlxx@163.com (F.L.); lichao8849@163.com (C.L.)
2 Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing, Xi’an 710121, China
3 Xi’an Key Laboratory of Big Data and Intelligent Computing, Xi’an 710121, China
4 College of Economics and Management, Xi’an University of Posts & Telecommunications, Xi’an 710121, China
* Correspondence: tuo_sh@126.com (S.T.); hehongtbw@126.com (H.H.)
Abstract: To accurately predict the economic development of each industry in different types of
regions, a deep convolutional neural network model was designed for predicting the annual GDP;
GDP growth index; and primary, secondary and tertiary industry growth values of each. In the model,
raw industrial data are preprocessed by a normalization operation and subsequently transformed
by the BoxCox method to approach the normal distribution. Panel data of consecutive years are
constructed and used as input to the deep convolutional neural network, and industrial data of year
t + 1 are used as the output of the network. Simulation experiments were conducted to analyze
23 years of industrial economic data from 31 provinces, municipalities, and autonomous regions in
China. The experimental results show that R-squared value is larger than 0.91 for all 31 provinces
and root mean squared log errors (RMSLE) of all regions are less than 0.1, which demonstrate that the
proposed method achieves high prediction accuracy with generalization capability and can accurately
predict the economic growth trends of different types of regions.
Keywords: deep convolutional neural network; regional economy; industrial economic big data
1. Introduction
In recent years, political issues, such as trade protectionism, racism, and populism,
have brought great uncertainty to economic and social development worldwide. In partic-
ular, the trade dispute between China and the United States has produced major challenges
to world economic development. In the current complex international environment, China,
as the world’s largest developing country, must urgently revise its thinking on economic
development by expanding domestic demand, adjusting its industrial structure, and estab-
lishing a new strategy for regional economic development with adaptive capabilities. In the
19th Party Congress report, Xi Jinping clearly pointed out that China has entered a new era.
He also wants to build China into a harmonious and beautiful modern socialist country,
and to achieve sustainable, coordinated development of China’s domestic economy by the
middle of this century [1]. In 2020, China proposed the “Double Cycle: A New Pattern
of National Economic Development”, which featured a major domestic economic cycle
and dual domestic and international economic cycle promoting each other. The proposal
focuses on strengthening the coordinated development of the domestic economy and
enhancing the sustainable development of China’s economic grand cycle by enhancing the
sustainable development of China’s regional industrial economies [2]. One of the main rea-
sons for the establishment of the European Union (EU) is to strengthen economic relations
between European countries, aiming at enhancing the external economic competitiveness
of the EU and the influence of the EU countries in the world through the cooperation of
the EU economies. In 2021, President Biden signed a trillion-dollar stimulus package to
https://www.mdpi.com/journal/sustainability
Sustainability 2021, 13, 12789 2 of 11

strengthen U.S. infrastructure and manufacturing, designed to promote circular and rapid
growth in the U.S. domestic economy.
However, due to current complex international political relations and dynamic macroe-
conomic development, methods to accurately predict the future development of regional
industrial economies based on current economic status and understand the interactions
between different industrial systems and their impact on regional economic development
have all become important issues in selecting regional pillar industries, enhancing the
competitiveness of cities and the synergistic development of regional economies. By mining
historical data on the economic development of each region in the country, this study in-
vestigates the key economic industries developed in each region and analyzes and predicts
the growth potential of regional industrial economic development to establish a theoretical
basis and value reference for the selection of advantageous industries in the region and the
coordinated development of each region.
At present, regional and industrial economy research is rich and wide-ranging and
involves relatively mature research methods. Scholars carrying out multilevel and multi-
perspective studies on different research topics using various research methods have
produced a steady stream of research results. In terms of research content, most of the
literature focuses on specific industries, such as agriculture [3], manufacturing [4,5], real
estate [6], logistics [7], finance [8], productive services [9], cultural industries [10], informa-
tion industries [11], and other economic industries. The association of regional industries
and their ripple effects [12] have also become the focus of research in recent years, and
several rationalized policy recommendations have been proposed to promote industrial
structure upgrading [13]. We introduced a multiregional economic correlation analysis
method based on big data [14]. These recommendations are powerful tools for judging the
economic contribution of industries and selecting leading regional industries. Tianren Yang
et al. investigated the residential mobility patterns of households living in low-income
neighborhoods and studied the neighborhood characteristics that influence their mobil-
ity [15]. Lingqiang Kong et al. summarized the big data-based urban environment, society,
and sustainability (UESS) research using a systematic review approach in combination with
bibliometric and thematic analyses [16]. Zaheer Allam and Zaynah A. Dhunny proposed a
new framework binding AI technology and cities while ensuring the integration of key
dimensions of culture, metabolism, and governance [17].
In terms of research methods, various economic forecasting models have been pro-
posed to predict the future development of relevant industries based on the current state of
the industrial economy. Tingting Zhang et al. proposed an FWA-SVR forecasting model to
forecast the tourism economy [18]. Cheng Mao-Lin et al. used the gray model GM (1,1)
to predict the growth trends of China’s real estate economy [19]. In another study [20],
China’s macroeconomic system was analyzed by constructing a BMF-VAR model to fore-
cast China’s macroeconomy. In [21], a deep-learning LSTM model used to forecast and
analyze the inflation rate in China achieved good performance. Ji Yao used commod-
ity price big data to construct a MIDAS model for analyzing and forecasting CPI and
PPI [19,22]. Xia Maosen used a deep neural network CNN-LSTM model to analyze and
forecast the Chinese consumer confidence index [23]. Pichayakone Rakpho et al. proposed
a Bayesian vector autoregressive (BVAR) model to forecast energy demand and supply
more accurately [24]. Yifei Lyu et al. constructed a dataset of 77 countries representing
over 90 percent of global GDP and forecasted US economic growth in downturns using
cross-country data [25]. Oscar Claveria et al. presented a machine-learning method for
sentiment indicators construction and economic forecasting with evolved confidence indi-
cators [26]. To enhance the prediction accuracy of COVID-19 and strengthen the economic
management and control, Xuan Tang et al. introduced a self-correcting intelligent pan-
demic prediction model [27]. To accurately forecast business failure prediction, Soo Young
Kim et al. developed three prediction models (entire period model, economic downturn
model, and economic expansion model) using WEKA3.9 [28].
 Sustainability 2021, 13, x FO
Sustainability 2021, 13, x FOR PEER REVIEW 3 of 11
Sustainability
Sustainability
Sustainability 2021,
2021,
2021,
Sustainability 13,
13,
2021, x xFOR
13,
13,xFOR
xFORPEER
PEER
PEER
FOR REVIEW
REVIEW
REVIEW
PEER REVIEW 3 33of
3of1111
ofof 11
11

Sustainability 2021, 13, 12789 3 of 11

Kim
Kim
KimKimetet Kim
etetal.
al.al. et al. developed
developed
developed
developed
al. developed three
threethree
three three
prediction
prediction
prediction
predictionpredictionmodels
models
models
models models
(entire
(entire
(entire
(entire (entire
period
period
period
period period
model,
model,
model,
model, model,
economic
economic
economic
economic economic
downturn
downturn
downturn
downturn downturn
model,
model,
model,
model, model,
and
andand
and and
economic
economic
economic
economic economic expansion
expansion
expansion
expansionexpansion model)
model)
model)
model) model)
using
using
using
using using
WEKA3.9
WEKA3.9
WEKA3.9
WEKA3.9 WEKA3.9 [28].
[28].
[28].[28]. [28].
The
TheThe
Theabove Theforecasting
above
above
above above
forecasting
forecasting
forecasting forecasting
models
models
models
models models
have
havehave
have have
achieved
achieved
achieved
achieved achieved
sound
sound
sound sound sound
forecasting
forecasting
forecasting forecasting
forecasting results
results
results
results results
ininin relatedin related
related
related
in related eco- eco- eco-
eco-
eco-
nomic The above
fields, but forecasting
the current models
research have achieved
remains focusedsound on forecasting
macroeconomic resultsforecasting
in related eco-
nomic
nomic
nomic
nomic fields,
fields,
fields,
fields, but
butbut
butthethe
the
the current
current
current
current research
research
research
research remains
remains
remains
remains focused
focused
focused
focused on on
on macroeconomic
macroeconomic
macroeconomic
on macroeconomic forecasting
forecasting
forecasting
forecasting and and and
and
and
nomic
analysis fields, but
of a economic
single the current
economic research
industry. remains Since focused
regional on macroeconomic
economic development forecasting and
is influ-
analysis
analysis
analysis
analysis ofof
ofaaasingle
of asingle
single
single economic
economic
economic industry.
industry.
industry.
industry. Since
Since
Since
Since regional
regional
regional
regional economic
economic
economic
economic development
development
development
development isisisis
influ-
influ-
influ-
influ-
analysis
enced of a
by multiple single economic
factors, industry.
multiregional, Since regional
multifactor, economic
and related development industry is influenced
correlation
enced
enced
enced
enced by by
byby multiple
multiple
multiple
multiple factors,
factors,
factors,
factors, multiregional,
multiregional,
multiregional,
multiregional, multifactor,
multifactor,
multifactor,
multifactor, and
andand
and related
related
related
related industry
industry
industry
industry correlation
correlation
correlation
correlation
by multiple
analysis factors, multiregional,
forecasting is obviously multifactor,To
insufficient. and relatedthe
consider industry
correlation correlationof analysis
multiple in-
analysis
analysis
analysis
analysis forecasting
forecasting
forecasting
forecasting isisisobviously
isobviously
obviously
obviously insufficient.
insufficient.
insufficient.
insufficient. ToTo
To consider
consider
consider
To consider the
thethe
the correlation
correlation
correlation
correlation ofof
of multiple
multiple
multiple
of multiple in-
in-in-
in- and
forecasting
dustries and is obviously
factors insufficient.
simultaneously, Towe consider
construct the a correlation
multilayer of multiple
convolutional industries
neural net-
dustries
dustries
dustries
dustries and
and
factorsand
and factors
factors
factors
factors simultaneously,
simultaneously,
simultaneously,
simultaneously, simultaneously, we constructwewe
wewe construct
construct
construct
construct
adata aaamultilayer
multilayer amultilayer
multilayer
multilayer convolutional
convolutional
convolutional
convolutional convolutional neural neural
neural
neural
neural
network net-
net-
net-net-
(CNN)
work
work
work
work work
(CNN)
(CNN)
(CNN)(CNN) (CNN)
toto
to build
build
build
to buildto
panelbuild
panel
panel
panel panel
economic
economic
economic
economic economicdata
data
data data with
with
with with with
temporal
temporal
temporal
temporal temporal
and
andand
and and
continuous
continuous
continuous
continuouscontinuous effects
effects
effects
effects effects
based
based
based
based onbased
on
onon on
to build
historical panel
data economic
from multiple data with
economic temporal
indicators and continuous
inregion.
each region. effects
We use based on historical
the convolutional
historical
historical
historical
historical
data data
data
data data
from from
fromfrom
from multiple
multiple
multiple
multiple multiple economic
economic
economic
economic economic indicators
indicators
indicators
indicators
indicators ininin ineach
in
each
each each
each region.
region.
region.region. We
We We
WeWe use
use
use use
use the
the
the the convolutional
convolutional
convolutional
the convolutional
convolutional kernel
kernel
kernel
kernel
kernel kernel
ofofof this
this
this
of of
neural
this this
neural
neural
neural neural
network
network
network
network network
toto
toto mine
mine
mine mine to mine
correlationscorrelations
correlations
correlations
correlations ininin the
the
in the
the in
panelthe
panel
panelpanel panel
data
data
data data toto
todata to
accurately
accurately
accurately
to accuratelyaccurately
predict
predict
predict
predict predict
of this
theeconomic neural
future economicnetwork to
development mine correlations
trends of in the
the regions. panel data to
The contributions accurately of predict
our work the
are
the
the
thethefuture
future
future
future
future economic
economic
economic
economic development
development
development
development
development trends
trends
trends
trends
trendsofof
of the
the
of the
the
of regions.
regions.
regions.
regions.
the The
The
regions. The
The contributions
contributions
contributions
contributions
The contributions ofofof
our
of our
our
our work
work
work
of work
our are
areare
are are
work
summarized
summarized summarized
summarized
summarized asas
as follows:as
follows:
follows:
as follows: follows:
summarized as follows:
(1)
(1)(1)AAA
(1) (1)
novel
novel
novel
A novelA novel
multilayer
multilayer
multilayer
multilayer multilayer CNN
CNNCNN
CNN CNN
isis is constructed
isconstructed
isconstructed
constructed
constructed forfor
for
for for predicting
predicting
predicting
predicting
predicting the
thethe the economic
economic
economic
economic
the economic development
development
development development
development ofof
of
of of of
(1) A novel multilayer CNN is constructed for predicting the economic development
primary, primary,
secondary secondary and and
tertiary tertiary
industries industries
ofof of
multi-regions.multi-regions.
InIn the In
CNN,the CNN,
the GDP the GDP of
primary,
primary,
primary, secondary
secondary
primary,secondary secondary and
and and tertiary
tertiary
tertiary
and tertiary industries
industries
industries
industriesofof multi-regions.
multi-regions.
multi-regions.
of multi-regions. Inthe
In the
theCNN,
InCNN,
thethe
CNN, theGDP
the
CNN, GDPtheofof
GDP of
of of
GDP
aaaregion
aregion
region a is
region
isisseparated is separated
isseparated
separated from
fromfrom the from
the
the otherthe indicators
other
other other
indicators indicators
ofofof the
the of the before
region
region region
before
before before
full
full full connection
connection
connection
regiona region separated
is separated from the
from theindicators
other indicators
other indicators the
of theregion
of region
the region before full connection
full
before connection
full connection
layers,
layers,
layers,
layers, layers,
which
which
which
which which
aims
aims
aims aims totoaims
to enhance
enhance
to to enhance
enhance
enhance the
the the
the the predict
predict
predict
predict
predict accuracy
accuracy
accuracy
accuracy accuracy
according
according
according
according according
tototo the
the
to the
the to
GDP
GDP
GDP the
GDP GDP
value.
value.
value.
value. value.
layers, which aims to enhance the predict accuracy according to the GDP value.
(2)
(2)
(2)(2)ToTo
To(2)reduceTo
reduce
reduce reduce
the
the data
data the data heterogeneity
heterogeneity
heterogeneity among
among among
the
the the
economic
economic economic
indicators
indicators indicators
ofofof each
each of each
regions
regions regions
and
and and
To
(2) To the
reduce reduce data
the data
theheterogeneity
heterogeneity
data heterogeneity amongamong the
amongtheeconomic
economic
the economic indicators
indicators
indicators each
of each regions
of regions
each and and and
regions
the
thethe the
differences
differences
the differences
differences
differences inin
in the
in the
the in
economic
the the
economic
economic
economic economicdevelopment
development
development development
development levels
levels
levelslevels levels
of
of each
oflevels
of each
each eachof each
region,
region,
region,
region, region,
the
the
the the datathe
data
data data are
are data
are firstare
first
first
are first first
the differences in the economic development of each region, the data are first
normalized
normalized
normalized normalized
normalized tototo balance
balance
balance
to balanceto balance
thethe
the the
deviations
deviations
deviations
the deviationsdeviations
among
among
among
among among
the
the the
the the
indicators,
indicators,
indicators, indicators,
indicators, and
andand
and thenand
then
then then the
the then
the
the the
normalized
normalized
normalized
normalizednormalized
normalized to balance the deviations among the indicators, and then the normalized
data
data
data datawere data
were
were
were were
transformed
transformed
transformed
transformedtransformed exponentially
exponentially exponentially
exponentially
exponentially toto
to balance
balance
to to balance
balance
balance the
the the
the the deviations
deviations
deviations
deviations
deviations among
among
amongamong among
the
the the the regions.
regions.
regions.
regions.
the regions.
data were transformed exponentially to balance the deviations among the regions.

2.2.2. 2. Prediction
2. Prediction
2.Prediction
Prediction
Prediction
Prediction Models
Models
Models Models
Models
Models Based
Based
Based on
Based Based
Based
on
on onConvolutional
on
Deep
Deep
Deep
on DeepDeep
Deep Convolutional
Convolutional
Convolutional
Convolutional Neural
Neural
Neural
Convolutional Neural
Neural Networks
Networks
Networks
Networks
Networks
Neural Networks
2.1.
2.1.
2.1. 2.1.Architecture
2.1.
Model
Model
Model
2.1. Model Model Architecture
Architecture
Architecture
Architecture
Architecture
As
AsAs
As As shown
As
shown
shown
shown
shown shown
inin
in
in in Figure
in
Figure
Figure
Figure
Figure Figure
1,1,our
1, 1,our
our
our 1,
1,
deepour
our
deep
deepdeep deep
deep convolutional
convolutional
convolutional
convolutional
convolutional
convolutional neural
neural
neural
neural neural
neural
network
network
network
networknetwork
network
[29]
[29]
[29] [29] inputs
[29]
inputs
inputs
inputs
[29] inputs inputs
the
thethe the gross
the
gross
gross
gross
the gross gross
regionalregional
regional
regional
regional
regional product
product
product product
product
product (GDP),
(GDP),
(GDP),
(GDP), (GDP),
(GDP),
GDP
GDP
GDP GDP GDP
GDP
growth
growth
growth
growth growth
growth
index,
index,
index,
index, index,
index,
primary
primary
primary
primaryprimary
primary
industry
industry
industry industry
industry
industry growth
growth
growth
growth growth
growth
index,
index,
index, index,
index,
secondary
secondary
secondary
index, secondary
secondary
secondary
industryindustry
industry
industry
industry
industry growth
growth
growth
growthgrowth
growth
value,
value,
value,
value, value,
value,
and
andand
and and tertiary
and
tertiary
tertiary
tertiary tertiary
tertiary industry
industry
industry
industryindustry
industry
growth
growth
growth
growth growth
growth
value
value
valuevalue value
value
for
forfor
for for
for nn consecutive
nnnconsecutiveconsecutive
nconsecutive
consecutive
consecutive years
years
yearsyears years
inyears
in
in
in in aa
aaaa in
region
region
region
regionregion
region
asasas
as as
as
panel
panel
panel
panel panel
panel
data
data
datadata data
data
into
into
into the
the
into theinto
into
the the
the
network
network
network
networknetwork
network
tototo to to
predict
predict
to predict predict
predict
predict the
the
the the the
output
output
output
the output output
output primary,
primary,
primary,
primary,
primary,
primary, secondary,
secondary,
secondary,
secondary,
secondary,
secondary, and
and and
and
and ter- ter-
and
tertiary
ter-
ter-
ter-
tiary
tiary
tiary
tiary industry
tiary
industry
industry
industry
industry growth
industry
growth
growth
growth
growth values
growth
values
values
values
values forfor
values
for
for
the
for the
the
the region
for
regionthe in
region
region
the region in
region
in
in year
year
year
year
in yearnnn
inn ++
year
+n 1.1.n + 1.
++1.1.
1.

Figure 1. Structur
Figure Figure
Figure
1.1. 1. Structure
1. Structure of
of the economic
the economic forecasting
forecasting model
model based
based aon on a
a deep deep convolutional
convolutional neural
neural network.
network. The The circle
circle points
points ,
Figure
Figure
Figure 1. 1.Structure
Structure
Structure
Structure ofof
of the
the
of the economic
economic
economic
the economic forecasting
forecasting
forecasting
forecasting model
model
model
model based
based
based on
based onaadeep
on
on adeep
deep
deep convolutional
convolutional
convolutional
convolutional neural
neural
neural network.
network.
network.
neural The
The
network. The circle
circle
circle
The points
points
points
circle points , ,
, , , and represent the values of GDP, GDP growth index, primary industry growth value, industry growth v
secondary
, , , , , , ,, , ,, , and andand represent
and
and represent
represent
represent the
representthe values
the
the values
values
values
the of
values GDP,
ofofofGDP,
of GDP,
GDP, GDP
GDP, GDP
GDP
GDP growth
GDP growth
growth
growth index,
growth index,primary
index,
index,
index, primaryindustry
primary
primary
primary growth
industry
industry
industry value,
growth
growth
growth
industry secondary
value,
value,
value,
growth secondary
secondary
secondary
value, industry
secondary
industry industry
growth growth
value, value,
and and tertiary
tertiary industry industry
growth growth
value, value, respectively,
respectively, of of one years.
industry
industry
industry growthgrowth
growth
growth value,
value,
value,value,and
and and
and tertiary
tertiary
tertiary
tertiary industry
industry
industry
industry growth
growth
growthgrowth value,
value,
value,value, respectively,
respectively,
respectively, ofof
respectively, oneofone
of
oneone years.
years.
years.
one
years.years.

The
TheThe
The Thedivides
The
model
model
model
model model
model
divides
divides
divides divides
divides
the
the the thedata
the
input
input
input
the input input
input
data
data
data data
data
into
into
into two
into into
into
two
two
two two parts.
two
parts.
parts.
parts.
parts. parts.
(1)
(1)(1)
(1) (1)
Regional
Regional
Regional
RegionalRegional
(1) Regional
GDPGDP
GDP
GDP GDP
GDP
network:
network:
network:
network:network:
network:
the
the the
the the
regional
regional
regional
regionalregional
the regional
GDP
GDPGDP
GDP GDP
GDP
data
data
data data are
are data
data
are are independent
are independent
independent
independent
independent
are independent ofof
of other
other
other
of of
of
other other
other
indicator
indicator
indicator indicator
indicator
indicator
data
data
data
data data
data
before
before
before
before before
before
the
thethe
the fully the
fully
fully fully fully
theconnected
fully
connected
connected
connectedconnected
connectedlayer
layer
layer layer layer
layer
ofofof the
the
of the of
of the
neural
neural
the neural neural
the neural
neural network,
network,
network,
network, network,
network,
and
and
andand and
and
feature
feature
feature
feature feature
feature information
information
information
information
information isisisis isis
information
extracted
extracted
extracted
extracted
extracted
extracted using
using
usingusing using
using
convolutionalconvolutional
convolutional
convolutional
convolutional
convolutional neural
neural
neural
neural neural
neural
networks
networks
networks
networksnetworks
networks
(CNNs),
(CNNs),
(CNNs),
(CNNs), (CNNs),
(CNNs),
batch
batch
batch
batch batch
batch normalization,
normalization,
normalization,
normalization,
normalization,
normalization, and
and and
and the
the and the
and
the
the the
ReLU
ReLU
ReLU
ReLU ReLU
ReLU
activationactivation
activation
activation activation
activation function.
function.
function.
function.function.
function.
This
This
ThisThis This
design
design
design
designThisdesigndesign
reflects
reflects
reflects reflects
reflects reflects
that
that
that the
the
that thethat the
that
regional
regional
regional
the regional regional
the
GDP
GDP
GDPGDP can GDP
regional
can
can can
GDP
mirror
mirror
mirror
can mirror mirror
the
the can
the the
mirror
overall
overall
overall
the overall overall
the
economiceconomic
overall
economic
economic
economic level
level
level
levelofoflevel
economic
of the
the
of the
the of the region,
level
region,
region,
region,
region, of the is
which
which
whichwhich which
region,
isis is
isequivalentequivalent
which
equivalent
equivalent
equivalent tois
to
to to gauging
equivalent
gauging
gauging
gauging
to gauging the
the to
the the regional
gauging
regional
regional
regional
the regional the
economic
economic
economiceconomic
regional
economic economic
devel-
devel-
devel-
devel- devel-
opment
opment
opment
opmentopment
development
level
level
level
levelby level
by
by
by by
regional
regional
regional
regional regional
levelGDP by
GDP
GDP
GDP totoGDP
regional
todivide
to divide
divide to
divideGDP divide
the
thetheto the
divide
regions
regions
regions
the regions regions
the
(similar
(similar
(similar
(similar(similar
regions
tototothe
to
thethe to the
(similar
role
role
role
the ofof
role ofrole
to
the
the
of the of
the the
role
panel
panel
panel
the panel panel
of
thresh-
thresh-
thresh-
thresh- thresh-
the panel
old
old
oldold old
model)
model)
model)
model) model)
thresholdand
andand
and usingand
model)
using
using using usingand
different
different
different different
differentusingforecasting
forecasting
forecasting forecasting
different
forecasting strategies
strategies strategies
forecasting
strategies
strategies for
forfor
for for
strategies
regions
regions
regions
regions regions
with for
with
withwith with
regions
different
different
different different
different with
develop-
develop-
develop-
develop-develop-
different
ment
ment
ment
ment ment
levels
levels
levels
levels levels
development
inininthe
inthe
the the in
fullythe
levels
fully
fully fully
in the
connected
connected
connected
fully connected
connected fully
layer.
layer.
layer.
layer. layer. layer.
connected
(2) Regional panel data network: The regional GDP growth index and the primary,
secondary, and tertiary industry growth values are combined to form panel data to identify
Sustainability 2021, 13, 12789 4 of 11

correlations between industries and the influence of industry data on later growth values
within n consecutive years using neural network convolution kernels. First, three CNN
layers (including convolution, batch normalization, ReLU activation function, and pooling
operations) are used to extract time series data features and potential industry connections.
The inputs of the fully connected layers are the output values of (1) and (2), and the
two fully connected layers integrate the outputs of the GDP and panel data networks to
further optimize the learning of the multichannel outputs of the two networks.

2.2. Model Analysis

(1) Convolution: The main purpose of the convolution layer is to extract features from
the input panel and sequence data using the following expression.

C kx ky  
zl +1 (i, j) = ∑∑∑ Z l s x i + x, sy j + y wcl +1 ( x + y) + b


(1)
c =1 x =1 y =1

where zl (i, j) denotes the value of the i-th row and j-th column data of the feature image
(panel or sequence) in the l-th layer, C represents the number of channels, (k x , k y ) indicates
the size of the convolutional kernel, (s x , sy ) is the convolution step size, wcl +1 represents the
weight matrix of channel c of l + 1, and b is a bias value.
(2) Batch Normalization (BA): BA can stabilize the training process of the network,
improve training speed, and minimize problems (such as gradient disappearance and
overfitting [30] using the following expression).

  zl(b) (i, j) − u(i, j)

yl(b) (i, j) = BN zl(b) (i, j) = γ· +β (2)
σ (i, j)

where zl(b) (i, j) denotes the input value of the b-th sample (Row i, Column j) in the current
batch. Here, zl(b) (i, j) denotes the output value of the convolution operation; u(i, j) and
σ (i, j) denote the mean and standard deviation of the batch sample at that position (Row i,
Column j), respectively; and parameters γ and β are used to control the mean and variance
of yl(b) , respectively.
(3) Relu: Relu is an activation function used to recalculate and activate the data after
normalization as follows. 
l x , x>0
y = x (3)
α ( e − 1) , x ≤ 0

where yl is the output value of the l-th layer after activation by the Relu function.
(4) Pooling: Down sampling, dimensionality reduction, and feature compression are
conducted on the output data of the convolutional operation. Maximum pooling is used in
the first and second convolutional layers, and the mean pooling operation is used in the
third convolutional layer.
(5) Loss function: The model uses the Huber loss function (as shown in Equation (4))
to optimize the prediction results as follows.
n
loss(y, y0 ) = n1 ∑ zi
  i =1  2 
 0.5 y − _ _

− (4)
y , i f y i < 1

 i _ i i
y
zi =
 y − y  − 0.5, otherwise
 
i i

_
where y and y denote the real and predicted values of the network, respectively.
(6) Optimization algorithm: For the loss function, the distance between the predicted
network output and true values is minimized by optimizing the network parameters.
The Adam optimization approach, which is solved iteratively by performing the gradi-
Sustainability 2021, 13, 12789 5 of 11

ent derivative of the loss function [31], is adopted for the optimization of the network
parameters. The Adam algorithm iterative formula is written as follows.
mt
θ t = θ t −1 − α √ (5)
vt + ξ

e t = β1 · m
m e t −1 + (1 − β 1 ) · g t
mt = me tt+1
1− β 1
vet = β 2 · vet−1 + (1 − β 2 ) · gt 2
vt = vett+1
1− β 2

where θt is the parameter value of the network at the t-th iteration, mt denotes the first
moment estimate, vt denotes the second moment estimate, and gt denotes the gradient
value at time t. α represents the learning rate during the iteration (0.01 in this paper); β 1
and β 2 both represent the exponential decay rate (β 1 is used to control the momentum of
the exponential shift and the weight of the gradient at generation t (0.9), respectively; and
β 2 is used to control the weight of the squared gradient at generation t − 1 (0.999)). ξ is a
very small positive number (1 × 10 − 8) designed to avoid the denominator of Equation (5)
being 0.

3. Analysis and Forecast of the Industrial Economy of 31 Provinces and Municipalities

in China
3.1. Dataset
In this paper, we use national macro annual data from the CEE statistical database
(https://db.cei.cn/, accessed on 20 May 2021), including the annual GDP; GDP growth
index; and growth values of primary, secondary, and tertiary industry economies for
31 provinces, municipalities directly under the central government, and autonomous
regions of China, from 1997 to 2020.
Data preprocessing: To enable uniform network training, all the data were first
normalized as follows.
j
j xi − min( xi )
xi = (6)
max( xi ) − min( xi )
j
where xi denotes the data corresponding to the i-th indicator; xi is the data value of the j-th
row in the i-th indicator; and max( xi ) and min( xi ) denote the maximum and minimum
values of the data in the i-th indicator, respectively.
Due to unbalanced economic development in the 31 given provinces in China, the
economic levels of the developed eastern regions and the northwestern provinces differ
greatly. Figure 2 (statistical histogram and QQ plot) shows the distribution of the data in
three industries (after normalization) for 31 provinces, municipalities, and autonomous
regions in China over 20 years. As seen in Figure 2, the data distribution of the three
industries has a very large deviation, which gravely affects the accuracy of the prediction
model. To eliminate the large variability of the data and convert it to an approximate
normal distribution, the normalized data are transformed using the following method.
q
_j 5 j
xi = xi (7)

The transformed data follow the approximate normal distribution of Figure 3, which
basically satisfies the conditions of regression analysis.
In this paper, which is based on regional economic data in China from 1997 to 2020,
the regional data of five consecutive years are used as the network input, X, and the data of
the sixth year are used as the output, Y. Each region generates 19 sample data (each sample
is a matrix panel data with 5 rows and 5 columns), yielding a total of 19 × 31 = 589 sample
data. Due to the small number of samples, the network is easily overfitted with respect to
prediction learning for the 31 regions. To improve generalization capability and enhance
Sustainability 2021, 13, 12789 6 of 11

Sustainability
Sustainability 2021,
2021, 13,
13, xx FOR
FOR PEER
PEER REVIEW
REVIEW 66 of
of 11
11

the prediction performance of the network, the data with 589 samples are randomly fine-
tuned to generate new sample data, resulting in a total of 2565 sample data points. During
The
The transformed
training, transformed
the samples aredata follow
follow the
datarandomly the approximate
approximate
shuffled, normal
normal
and the distribution
distribution
network of Figure
ofin
is trained Figure 3,
3, which
batches,which
with
basically
basically
100 satisfies
satisfies
samples the
the conditions
randomly conditions of regression
of each
selected for batch.analysis.
regression analysis.

QQ
QQPlot
PlotofofTertiary
Tertiaryindustry
industryData
Dataversus
versusStandard
StandardNormal
Normal
QQ
QQPlot
PlotofofSecondary
Secondaryindustry
industryData
Dataversus
versusStandard
StandardNormal
Normal 11
11

0.8
0.8 0.8
0.8
Sample
Input Sample

Sample
Input Sample

0.6
0.6
0.6
0.6
of Input

0.4
0.4
of Input

0.4
0.4
Quantiles of

Quantiles of

0.2
0.2
Quantiles
Quantiles

0.2
0.2
00

-0.2
-0.2 00

-0.4
-0.4
-4-4 -3-3 -2-2 -1-1 00 11 22 33 44 -0.2
-0.2
Standard
StandardNormal
NormalQuantiles
Quantiles -4-4 -3-3 -2-2 -1-1 00 11 22 33 44
Standard
StandardNormal
NormalQuantiles
Quantiles

Figure
Figure 2.
Figure 2. Histogram
2. Histogram and
Histogram and QQ
and QQ chart
QQ chart of
chart of the
of the distribution
the distribution of
distribution of the
of the data
the data for
for three
data for three industries.
three industries.
industries.

Figure
Figure 3.
Figure 3. Histogram
3. Histogram and
Histogram and QQ
and QQ chart
QQ chart of
chart of the
of the distribution
the distribution of
distribution of the
of the transformed
transformed data
the transformed data for
data for three
for three industries.
three industries.
industries.

3.2. Network Parameters

In
In this
this paper,
paper, which
which is is based
based onon regional
regional economic
economic data data inin China
China from
from 1997
1997 to to 2020,
2020,
the The network
the regional
regional data parameters
data of
of five are summarized
five consecutive
consecutive years
years are as Table
are used
used as 1. network
as the
the network input,
input, X,X, and
and the
the data
data
of
of theWe
the sixthimplemented
sixth year
year are the
are used
used as network
as the model
the output,
output, Y. using
Y. Each PyTorch,
Each region including
region generates
generates 19 data normalization,
19 sample
sample data
data (each
(each
normalization,
sample
sample is network
is aa matrix
matrix panel training,
panel data
data with 5prediction,
with 5 rows and 55and
rows and evaluation.
columns),
columns), yielding
yielding The source
aa total
total of code
of 19
19 31of== 589
×× 31 the
589
algorithm
sample
sample data. is available
data. Due
Due to
to theat https://github.com/shouhengtuo/Cnn_economy
the small
small number
number of of samples,
samples, thethe network
network is is easily (accessed
easily overfitted
overfitted with on
with
20 May
respect
respect to 2021).
to prediction
prediction learning
learning forfor the
the 31
31 regions.
regions. ToTo improve
improve generalization
generalization capability
capability and and
enhance
enhance the the prediction
prediction performance
performance of of the
the network,
network, the the data
data with
with 589589 samples
samples are are ran-
ran-
domly
domly fine-tuned
fine-tuned to to generate
generate new new sample
sample data,
data, resulting
resulting in in aa total
total ofof 2565
2565 sample
sample data data
Sustainability 2021, 13, 12789 7 of 11

Table 1. Deep CNN parameters.

Subnetwork 1 for Panel Data Subnetwork 2 for GDP Data

Number of Kernel Stride Padding Number of Kernel Padding
Layers Type Layers Type
Channels Size Size Size Channels Size Size
Convolution
Convolution 64 (3,1) (1,1) (2,1) Convolution 16 (3,1) (1,0)
layer 1
Convolution
Batch Batch
layer 1 64 / / / / 16 / /
normalization normalization
Relu 64 / / / / Relu 16 / /
Pooling 64 (3,1) / / / Pooling 16 (3,1) /
Full
Convolution 32 (1,3) (1,1) (1,2) connection Input size: 16; output size: 32
Convolution layer 0_0
layer 2 Full
Batch
32 / / / connection Input size: 32; output size: 16
normalization
layer 0_1
Relu 32 / / /
Pooling 32 (1,2) / /
Convolution 16 (2,2) (1,1) (0,0)
Convolution Batch
16 / / /
layer 3 normalization
Relu 16 / / /
Pooling 16 (2,2) / /
Merge the output of subnetwork 1 and subnetwork 2
Full connection layer 1_0 Input size: 16, output size:32
Input size: 32, output size:3
Full connection layer 1_0
The output is the economic growth values of the three industries in the 6th year.

3.3. Results and Analysis

For network training, 10-fold cross-validation was used, and the value of the loss
function was essentially stable after 300 iterations. The trained network was used to predict
the last three years of data for three industries in 31 provinces, municipalities that are
directly under the central government, and autonomous regions of China. The following
two evaluation indices were used to evaluate the prediction accuracy of the network.
(1) R-Squared
n
2
∑ (ŷi − yi )
SS i =1
R2 = 1 − residual = 1 − n (8)
SStotal 2
∑ ( y − yi )
i =1

where yi and ŷi represent the true output value and network prediction of the ith
sample, respectively; and yi represents the mean value of the true value in the region.
(2) Root Mean Squared Log Error RMSLE
s
1 n
n i∑
RMSLE = (log(ŷi + 1) − log(yi + 1))2 (9)
=1

where yi and ŷi represent the true output value and network prediction of the ith
sample, respectively.
The value of R-squared is in the interval [0,1], and the closer its value is to 1, the more
accurate the prediction result is. In contrast, the traditional mean squared error (MSE)
value is influenced by the range of values of the original data, which is not sizeable enough
to determine the degree of accuracy. The value of RMSLE is not greatly influenced by the
size of the original data compared to the root mean square error RMSE; thus, it is easier to
discern the prediction accuracy with RMSLE.
In Table S1 (see Supplementary), the prediction accuracies of the three industries
in 31 provinces are summarized by the deep multilayer convolutional neural network
proposed in this paper. Figure 4 shows plots of the predicted and true values for Beijing,
Shanghai, Shaanxi, Xinjiang, Heilongjiang, Guizhou, Tibet, Guangdong, Shanxi, and Zhe-
jiang Provinces from 2002 to 2020. From Figure 4 and Table S1, it can be seen that the
Sustainability 2021, 13, 12789 8 of 11

deep convolutional neural network prediction model proposed in this paper can accurately
predict the economic growth values of the three industries in all regions. The R2 values
of the network model for the 31 regions are all greater than 0.9. With the exception of
R2 = 0.915671 for the secondary industry in the Shaanxi Province, R2 = 0.972776 for the
primary industry in Tibet, and R2 = 0.938831 for the primary industry in the Zhejiang
Province, the R2 values for the three industries are all greater than 0.98. In addition, only
the RMSLE of the secondary industry in the Shanxi Province is 0.114741 (which has a
relatively large error); the prediction errors of all the other regions are less than 0.1. This
indicates that the method in this paper achieves high prediction accuracy and can be used
to forecast the economic growth of regions with large variability.
The above experimental results and analysis indicate that the proposed forecasting
model has high-precision for predicting multi-industries and multiple economic indicators
of a region. The model can be easily used to the future economic development trends of
the regions. However, it has some limitations, as listed below.
(1) The multilayer CNN requires a large sample size to train the network, but the histori-
cal economic data of each region is general not sufficient due to the lack and missing
of historical data.
(2) The CNN has high prediction and recognition ability when it has seen similar sample
data; however, its prediction ability is deficient for the data that has never been
seen before.
(3) The predication model proposed in this work only considers the GDP; GDP grow
index; and the growth value of primary, secondary, and tertiary industry as inputs,
and it can also predict the future growth values of primary, secondary, and tertiary
industries in a region. However, regional economic forecasting should consider the
development trends of more precise industrial economies (e.g., financial industry,
tourism, ecological environment, population structure, etc.).
Future works:
Sustainability 2021, 13, x FOR PEER REVIEW  9  of  12 
(1) Construction of a multiregional and multi-industrial structure economic index system.
 
The indicators of regional population structure, ecological environment, tourism,
logistics, service industry, agriculture and industry are integrated to build a more
(2) scientific and complete
Construction  multiregional
of  correlation  andmodel 
relationship  multi-industry structure indicator
for  the  multiregional  system.
and  multi‐in‐
(2) Construction of
dustry structure.  correlation relationship model for the multiregional and multi-
(3) industry structure.
Prediction model construction for multiregional and multi‐industrial structure eco‐
(3) Prediction model construction for multiregional and multi-industrial structure eco-
nomic indexes. 
nomic indexes.
     

   
     
Figure 4. Cont.
     

   
     
Sustainability 2021, 13, 12789 9 of 11
   
     
     

   
     

     
   

 
   
   

 
 
 
Figure 4. Fitted curve of predicted and true values.

4. Conclusions
The accurate prediction of regional industrial economic development trends is crucial
for future economic decisions and sustainable coordinated development. To accurately
predict and analyze the economic development of three industries in each province, munici-
palities directly under the central government, and autonomous region of China, this paper
proposes a multilayer convolutional neural network prediction model that uses regional
historical economic data to train and learn a neural network. First, the historical data of
the 31 provinces are integrated, and the historical data of each of the three industries are
normalized and preprocessed. Due to large differences in the economic development levels
of each region, the data contain obvious bias characteristics that affected the prediction
accuracy of the neural network. For this reason, the data are first normalized to balance
the deviations among the indicators, and then the normalized data are transformed using
Equation (7) to balance the deviations among the regions. Second, by constructing panel
data and generating sample data, the sample size is increased to improve the generalization
Sustainability 2021, 13, 12789 10 of 11

capability of the network. The experimental results show that the proposed multilayer
convolutional neural network can accurately predict the economic development of each
region. One disadvantage of our method is that we only analyze and predict from the
combined data of GDP of primary, secondary, and tertiary industries, which cannot reflect
the development trends of each industry. Since the industries in the regions tend to influ-
ence each other and have complex correlations with each other, it would be more useful
to analyze and forecast from the perspective of industry big data. As future research, we
intend to collect the historical data of each industry in the regions and form more complete
regional economic big data by analyzing and integrating the data of each industry, i.e.,
population structure, education, agriculture, industry, service industry, finance, science
and technology, ecological environment, traffic conditions, water resource distribution, and
energy conditions, and then mine their correlations to construct a deep neural network
model. In addition, we will build a deep neural network model to forecast the development
of the all industries taken together.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10

.3390/su132212789/s1, Table S1: Prediction accuracy in 31 regions of China (R2,RMSLE).
Author Contributions: Investigation, T.C. and H.H.; Methodology, S.T.; Software, S.T.; Writing–
original draft, S.T., T.C., Z.F., Y.Z., F.L. and C.L.; Writing–review & editing, T.C. and H.H. All authors
have read and agreed to the published version of the manuscript.
Funding: This research was supported by the Ministry of education of Humanities and Social Science
project, China under Nos. 19YJCZH148, special project of Shaanxi Province for key disciplines in
higher education institutions and supported in part by major scientific research project of Shaanxi
Provincial Department of Education (No. 20JZ089).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The dataset is download from http://data.drcnet.com.cn/ (accessed
on 20 June 2020).
Acknowledgments: The author would like to thank all the editors, reviewers and referees for their
constructive comments.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Lin, Y. Economics in the New Era of China. Theory Stud. Explor. 2020, 1, 9–12.
2. Lin, Y. The New Dual Cycle Development Apporach. China Bond 2021, 3, 7–9.
3. Ye, L.; Li, Y.; Liang, W.; Zhang, H. Correlation and Ripple Effect of Agriculture Sector in Hainan Province. Chin. J. Trop. Agric.
2017, 37, 82–87.
4. Chen, X.; Zhu, J.; Li, G.P. Regional Disparity and Influencing Factors of Industrial Co-Agglomeration in Chinese Manufacturing.
Econ. Geogr. 2018, 38, 104–110.
5. Chen, X.; Zhu, J. The effect of manufacturing industry affiliation on inter-industry synergistic agglomeration. Soc. Sci. Front. 2018,
2, 257–261.
6. He, D.; Hu, X. A Study of Correlation Effect and Ripple Effect of Real Estate Industry Based on the Positive Analysis of
Input-output Tables in China. Rev. Ind. Econ. 2015, 2, 17–27.
7. Wei, M.; Wang, L.; Li, Y. The Industrial Association and Spreading Effects Resulting From Modern Logistics Industry. J. Bus. Econ.
2009, 12, 15–21.
8. Hu, R.; Huang, X.; Chen, L. Dynamic Effects of Financial Services Industry on the Optimization of Industrial Structure in Hunan.
Theory Pract. Financ. Econ. 2012, 33, 22–26.
9. Chen, J.; Wang, Y. Industry Linkages and Spatial Distribution of Producer Services—An Analysis Based on Urban Panel Data
Spatial Econometrics. Econ. Surv. 2016, 33, 96–101.
10. Li, J.; Wan, C. The Empirical Research on the Influential Factors of Shanghai Cultural and Creative Industries Development-Based
on the Gray Connection Model Analyzes. Comment. Cult. Ind. China 2017, 24, 202–219.
11. Zhou, M. Study on the correlation effect of information industry in Zhejiang Province, China. Res. Financ. Econ. Issues 2016, S1,
18–22.
Sustainability 2021, 13, 12789 11 of 11

12. Zhu, L.; Wang, M. Evolution Analysis of Industry Correlation and Spreading Effect of Circulation Industry Based on the
Inout-Outout Table. China Bus. Mark. 2018, 32, 75–84.
13. Wang, S. A study of industrial linkages and industrial ripple effects of the financial industry in Zhejiang Province, China. Manag.
Adm. 2014, 12, 86–88.
14. Tuo, S.; He, H. A Study of Multiregional Economic Correlation Analysis Based on Big Data—Taking the Regional Economy of
Cities in Shaanxi Province, China, as an Example. Sustainability 2021, 13, 5121. [CrossRef]
15. Yang, T.; Pan, H.; Zhang, X.; Greenlee, A.; Deal, B. How neighborhood conditions and policy incentives affect relocation outcomes
of households from low-income neighborhoods—Evidence from intra-city movement trajectories. Cities 2021, 119, 103415.
[CrossRef]
16. Kong, L.; Liu, Z.; Wu, J. A systematic review of big data-based urban sustainability research: State-of-the-science and future
directions. J. Clean. Prod. 2020, 273, 123142. [CrossRef]
17. Allam, Z.; Dhunny, Z.A. On big data, artificial intelligence and smart cities. Cities 2019, 89, 80–91. [CrossRef]
18. Tang, Z.; Wang, M.; Wei, D.; Liu, Z. Seasonally-Adjusted FWA-SVR Model and Its Application in Tourism Economic Forecast. J.
Syst. Sci. Math. Sci. 2021, 41, 1572.
19. Cheng, M.; Han, Y. Grey Model GM(1,1) Based on Variable Generation Coefficient and Its Applications. Stat. Decis. 2020, 36,
15–18.
20. Tang, C.; Ye, L. On China’s Macroeconomic Forecast Based on Bayesian Mixing Model. New Financ. 2021, 2, 24–30.
21. Chen, Y.; Liu, L.; Chen, X. Research on Inflation Rate Forecasting Based on Deep Learning LSTM Model. Res. Financ. Econ.
Issues 2021, 1–14. Available online: https://kns.cnki.net/KXReader/Detail?invoice=mzlTo0TGfkyB3xEJ0UHfGjP%2Fkd6
nU0C3aIOecU3%2B2PuwSNb%2BuUxo%2BfyA3T0D2ehPzg3Gs%2FjseCKwUKUKWM%2BhIm%2Fa7MSWw16%2BfWt%
2FwikgGTYZ9XADbi%2BcCwvpRa2o6JaB7sVejQayjH0iZrH4AFiwLxK4vbRpWS8BougEZB%2BjG5g%3D&DBCODE=CAPJ&
FileName=CJWT20210406000&TABLEName=capjlast&nonce=F519E1B03A874B58AA9FE624BBA0FC55&TIMESTAMP=16
37207398092&UID=WEEvREcwSlJHSldTTEYzVnBFbFJWVUp0Y2dHZmxNMjkyNHRDb3M0WnNSND0%3d%249A4hF_
YAuvQ5obgVAqNKPCYcEjKensW4IQMovwHtwkF4VYPoHbKxJw!!&PlatForm=kdoc (accessed on 20 May 2021).
22. Ji, Y. Big Data-based Mixed Frequency Macroeconomic Prediction and Monitoring Index Construction. Stat. Decis. 2021, 7, 36–39.
23. Xia, M.; Jiang, L. China’s Consumer Confidence Index Forecast Based on Deep Network CNN-LSTM Model. Stat. Decis. 2021, 7,
21–26.
24. Rakpho, P.; Yamaka, W. The forecasting power of economic policy uncertainty for energy demand and supply. Energy Rep. 2021,
7, 338–343. [CrossRef]
25. Lyu, Y.; Nie, J.; Yang, S.K.X. Forecasting US economic growth in downturns using cross-country data. Econ. Lett. 2021, 198, 109668.
[CrossRef]
26. Claveria, O.; Monte, E.; Torra, S. Economic forecasting with evolved confidence indicators. Econ. Model. 2020, 93, 576–585.
[CrossRef]
27. Tang, X.; Li, Z.; Hu, X.; Xu, Z.; Peng, L. Self-correcting error-based prediction model for the COVID-19 pandemic and analysis of
economic impacts. Sustain. Cities Soc. 2021, 74, 103219. [CrossRef]
28. Kim, S.Y.; Upneja, A. Majority voting ensemble with a decision trees for business failure prediction during economic downturns.
J. Innov. Knowl. 2021, 6, 112–123. [CrossRef]
29. Goodfellow, I.; Bengio, Y.; Courville, A. Deep Learning; MIT Press: Cambridge, UK, 2016; Volume 1, pp. 326–366.
30. Ioffe, S.; Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proceedings
of the 32nd International Conference on International Conference on Machine Learning (ICML’15), Lille, France, 7–9 July 2015;
JMLR.org: Boston, MA, USA, 2015; Volume 37, pp. 448–456.
31. Kingma, D.; Ba, J. Adam: A Method for Stochastic Optimization. In Proceedings of the 3rd International Conference on Learning
Representations (ICLR 2015), San Diego, CA, USA, 7–9 May 2015.