The
problem
of
demand
forecasting
for
e-commerce
retail
merchants
is
a
common
challenge
in
the
industry.
key
lies
how
to
accurately
predict
customer
and
solve
practical
problems
based
on
prediction
results.
research
this
topic
facilitates
platforms
adjust
their
inventory
timely
manner
so
as
not
only
satisfy
but
also
effectively
reduce
costs
operating
costs.
It
line
with
development
trend
industry,
such
intelligence,
personalization
customization,
integration
online
offline.
In
paper,
hybrid
ARIMA-LR
model
first
used
forecast
different
storage
sites
e-commerce.
This
combines
an
autoregressive
sliding
average
(ARIMA)
linear
regression
(LR)
improve
accuracy
stability.
Then,
genetic
algorithm
select
best
classification
metrics.
performance
metrics,
most
suitable
metrics
from
them
classify
time
series.
By
classifying
series
into
categories,
patterns
features
same
type
can
be
better
understood
analyzed.
helps
extract
useful
information,
identify
potential
trends
patterns,
provide
decision
support
management
warehouse
sites.
Grey Systems Theory and Application,
Год журнала:
2025,
Номер
unknown
Опубликована: Май 8, 2025
Purpose
This
study
aims
to
address
the
interaction
among
influencing
factors
in
real
systems
and
varying
intensity
of
impact
that
independent
variables
have
on
dependent
variables,
a
new
IDFGM(1,N,ri)
model
is
proposed.
Design/methodology/approach
Firstly,
grey
relational
analysis
utilized
screen
sequences
identify
their
interactions.
Secondly,
particle
swarm
optimization
employed
for
differential
orders
nonlinear
parameters
each
variable,
while
least
squares
method
used
calculate
structural
parameter
matrix,
constructing
time
response
function
model.
Finally,
applied
simulate
predict
carbon
dioxide
emissions
China
compared
with
other
models.
Findings
The
results
show
has
good
simulating
predicting
performance,
verifying
its
effectiveness.
newly
introduced
demonstrates
high
degree
compatibility
can
be
seamlessly
integrated
conventional
In
case
analysis,
shows
enhanced
predictive
performance
relative
benchmark
finding
suggests
articulated
this
successfully
captures
attributes
sequence
by
employing
order,
facilitated
algorithm.
Practical
implications
article
presents
scientifically
grounded
effective
forecasting
emissions.
outcomes
these
predictions
serve
as
theoretical
foundation
development
policies
aimed
at
reduction
energy
transition.
Originality/value
unique
contribution
incorporation
interactions
into
multivariable
prediction
models,
along
cumulative
both
account
variations.
Furthermore,
application
enabled
adapt
dynamically.
Fractal and Fractional,
Год журнала:
2024,
Номер
8(7), С. 396 - 396
Опубликована: Июль 2, 2024
The
fractional-order
grey
prediction
model
is
widely
recognized
for
its
performance
in
time
series
tasks
with
small
sample
characteristics.
However,
parameter-estimation
method,
namely
the
least
squares
limits
predictive
of
and
requires
to
address
ill-conditioning
system.
To
these
issues,
this
paper
proposes
a
novel
parameter-acquisition
method
treating
structural
parameters
as
hyperparameters,
obtained
through
marine
predators
optimization
algorithm.
experimental
analysis
on
three
datasets
validate
effectiveness
proposed
paper.
Frontiers in Energy Research,
Год журнала:
2023,
Номер
11
Опубликована: Дек. 14, 2023
As
a
clean
fossil
energy
source,
natural
gas
plays
crucial
role
in
the
global
transition.
Forecasting
prices
is
an
important
area
of
research.
This
paper
aims
at
developing
novel
hybrid
model
that
contributes
to
prediction
prices.
We
develop
combines
“Decomposition
Algorithm”
(CEEMDAN),
“Ensemble
(Bagging),
“Optimization
(HHO),
and
“Forecasting
model”
(SVR).
The
used
for
monthly
Henry
Hub
forecasting.
To
avoid
problem
data
leakage
caused
by
decomposing
whole
time
series,
we
propose
rolling
decomposition
algorithm.
In
addition,
analyzed
factors
affecting
multivariate
Experimental
results
indicate
proposed
more
effective
than
traditional
predicting
Information Technology And Control,
Год журнала:
2023,
Номер
52(4), С. 849 - 866
Опубликована: Дек. 22, 2023
Landslides
significantly
impact
economic
development
and
public
safety.
Aiming
at
the
problem
of
insufficient
prediction
accuracy
displacement
data
series
traditional
grey
Verhulst
model,
this
paper
proposes
a
fractional
model
optimized
using
beetle
tentacle
search
algorithm.
First,
based
on
order
operator
is
introduced
to
accurately
adjust
magnitude
between
cumulative
values,
constructing
order-based
model.
Expanding
accumulative
range
improves
performance.
Second,
optimized.
The
antennae
algorithm
finds
optimal
0
1
in
minimizing
average
relative
error.
Finally,
Heifangtai
landslide
group
from
Gansu
Province,
simulation
experiments
verified
that
has
higher
fitting
effect
than
Huang's
improved
GM
(1,1)
cubic
exponential
smoothing
DGM
(2,1)
error
2.949
%.
Results
show
data.
more
suitable
for
predicting
deformation.
系统科学与信息学报(英文),
Год журнала:
2024,
Номер
12(2), С. 245 - 273
Опубликована: Янв. 1, 2024
This
study
considers
a
nonlinear
grey
Bernoulli
forecasting
model
with
conformable
fractional-order
accumulation,
abbreviated
as
CFNGBM$(1,
1,
\lambda)$,
to
the
gross
regional
product
in
Cheng-Yu
area.
The
new
contains
three
parameters,
power
exponent
$\gamma$,
id="M3">$\alpha$
and
background
value
id="M4">$\lambda$,
which
increase
adjustability
flexibility
of
id="M5">$(1,
\lambda)$
model.
Nonlinear
parameters
are
determined
by
moth
flame
optimization
algorithm,
minimizes
mean
absolute
prediction
percentage
error.
id="M6">$(1,
is
applied
16
cities
area,
Chongqing,
Chengdu,
Mianyang,
Leshan,
Zigong,
Deyang,
Meishan,
Luzhou,
Suining,
Neijiang,
Nanchong,
Guang'an,
Yibin,
Ya'an,
Dazhou
Ziyang.
With
data
from
2013
2021,
several
models
established
results
show
that
has
higher
accuracy
most
cases.
Kybernetes,
Год журнала:
2024,
Номер
53(13), С. 72 - 100
Опубликована: Ноя. 27, 2024
Purpose
As
social
networks
have
developed
to
be
a
ubiquitous
platform
of
public
opinion
spreading,
it
becomes
more
and
crucial
for
maintaining
security
stability
by
accurately
predicting
various
trends
dissemination
in
networks.
Considering
the
fact
that
online
is
dynamic
process
full
uncertainty
complexity,
this
study
establishes
novel
conformable
fractional
discrete
grey
model
with
linear
time-varying
parameters,
namely
CFTDGM(1,1)
model,
accurate
prediction
trends.
Design/methodology/approach
First,
accumulation
difference
operators
are
employed
build
enhancing
traditional
integer-order
parameters.
Then,
improve
forecasting
accuracy,
base
value
correction
term
introduced
optimize
iterative
model.
Next,
differential
evolution
algorithm
selected
determine
optimal
order
proposed
through
comparison
whale
optimization
particle
swarm
algorithm.
The
least
squares
method
utilized
estimate
parameter
values
In
addition,
effectiveness
tested
event
about
“IG
team
winning
championship”.
Finally,
we
conduct
empirical
analysis
on
two
hot
events
regarding
“Chengdu
toddler
mauled
Rottweiler”
“Mayday
band
suspected
lip-syncing,”
further
assess
ability
applicability
seven
other
existing
models.
Findings
test
case
recent
reveal
outperforms
most
models
terms
performance.
Therefore,
chosen
forecast
development
these
events.
results
indicate
attention
both
will
decline
slowly
over
next
three
days.
Originality/value
A
help
has
higher
accuracy
feasibility
trend
prediction.
The
problem
of
demand
forecasting
for
e-commerce
retail
merchants
is
a
common
challenge
in
the
industry.
key
lies
how
to
accurately
predict
customer
and
solve
practical
problems
based
on
prediction
results.
research
this
topic
facilitates
platforms
adjust
their
inventory
timely
manner
so
as
not
only
satisfy
but
also
effectively
reduce
costs
operating
costs.
It
line
with
development
trend
industry,
such
intelligence,
personalization
customization,
integration
online
offline.
In
paper,
hybrid
ARIMA-LR
model
first
used
forecast
different
storage
sites
e-commerce.
This
combines
an
autoregressive
sliding
average
(ARIMA)
linear
regression
(LR)
improve
accuracy
stability.
Then,
genetic
algorithm
select
best
classification
metrics.
performance
metrics,
most
suitable
metrics
from
them
classify
time
series.
By
classifying
series
into
categories,
patterns
features
same
type
can
be
better
understood
analyzed.
helps
extract
useful
information,
identify
potential
trends
patterns,
provide
decision
support
management
warehouse
sites.
