Electricity
demand
forecasting
is
of
great
significance
in
the
field
energy,
which
helps
rational
planning
and
management
electricity
resources.
The
aim
this
study
to
develop
an
model,
based
on
a
fuzzy
time
series
analysis
approach.
A
large-scale
dataset
containing
time,
actual
values,
forecast
data
provided
by
Transmission
System
Operator
(TSO)
used.
covers
development
evaluation
univariate
multivariate
models.
For
models,
we
implemented
HOFTS,
WHOFTS
PWFTS
results
show
that
model
performs
well
all
orders
clearly
outperforms
predictive
performance
TSO.
achieved
impressive
accuracy
MAPE
values
as
low
0.87%.
In
terms
MVFTS,
Weighted
FIG-FTS
models
were
applied,
making
full
use
partitioning
weight
assignment.
Although
these
failed
outperform
TSO
performance,
they
demonstrated
lower
errors
forecasting,
showing
advantages
dealing
with
complex
correlated
data.
Risks,
Год журнала:
2024,
Номер
12(11), С. 174 - 174
Опубликована: Ноя. 4, 2024
The
increasing
population
and
emerging
business
opportunities
have
led
to
a
rise
in
consumer
spending.
Consequently,
global
credit
card
companies,
including
banks
financial
institutions,
face
the
challenge
of
managing
associated
risks.
It
is
crucial
for
these
institutions
accurately
classify
customers
as
“good”
or
“bad”
minimize
capital
loss.
This
research
investigates
approaches
predicting
default
status
customer
via
application
various
machine-learning
models,
neural
networks,
logistic
regression,
AdaBoost,
XGBoost,
LightGBM.
Performance
metrics
such
accuracy,
precision,
recall,
F1
score,
ROC,
MCC
all
models
are
employed
compare
efficiency
algorithms.
results
indicate
that
XGBoost
outperforms
other
achieving
an
accuracy
99.4%.
outcomes
from
this
study
suggest
effective
risk
analysis
would
aid
informed
lending
decisions,
deep-learning
algorithms
has
significantly
improved
predictive
domain.
This
research
paper
examines
the
capability
of
fuzzy
time
collection
for
hyperspectral
photograph
classification.
Fuzzy
series
(FTS)
is
a
in
which
standards
are
used
to
model
styles
within
facts.
FTS
can
be
explain
complex
temporal
records,
and
as
consequence
making
it
possible
categorize
photographs
more
extraordinarily
accurately.,
this
look
proposes
an
optimization
method
primarily
based
on
genetic
seek
techniques.
The
algorithm
designed
discover
high-quality
parameters
that
yield
first-rate
type
accuracy.
efficacy
proposed
technique
evaluated
facts
set
with
extraordinary
experimental
scenarios.
results
test
display
enhance
accuracy
photo
classification
use
considerably.
Hence,
gives
promising
classify
snapshots
efficiently.
affords
optimized
machine
category.
device
consists
3
levels:
pre-processing,
version
creation,
optimization.
Throughout
pre-processing
level,
statistical
spectral
analyses
executed
acquire
applicable
attributes
developing
collection.
construction
degree
then
uses
bushy
extract
between-class
separability
type.
It
followed
utilizing
stage,
related
software
differential
evolution,
minimize
complexity
while
still
enhancing
has
been
correctly
carried
out
real-international
dataset
demonstrates
widespread
upgrades
class
over
existing
methods.
Forecasting
the
trajectory
of
time
series
is
notably
challenging,
primarily
attributed
to
intrinsic
non-linearities
and
continually
shifting
dynamics
present
in
financial
markets.
In
this
research
initiative,
we
adopt
a
pioneering
methodology
by
leveraging
capabilities
an
evolutionary
algorithm
named
Barnacle
Mating
Optimization
(BMO)
for
precision
adjustment
weights
biases
within
Artificial
Neural
Network
(ANN).
This
intricate
optimization
process
results
development
hybrid
model,
aptly
BMO+ANN.
We
put
BMO+ANN
test
utilizing
it
forecasting
closing
prices
two
widely
tracked
currency
exchange
rates.
order
provide
comprehensive
comparison,
also
train
same
ANN
model
using
DE
PSO
algorithms
resulting
competitive
models
such
as
DE+ANN
PSO+ANN
engage
them
task.
The
performance
evaluation
carried
out
RMSE
metric.
Remarkably,
conclusively
demonstrate
that
outshines
its
ability
make
more
accurate
forecasting,
underscoring
effectiveness
BMO
tackling
complexities
rate
forecasting.
Fuzzy
logic
has
come
to
be
a
crucial
tool
for
processing
and
interpreting
facts
in
diverse
fields,
which
includes
the
evaluation
type
of
time
collection
information.
This
paper
offers
application
fuzzy
series
category
networks.
First
off,
discusses
basics
common
sense,
include
ideas
units
sense
operations.
It
then
introduces
fuzzy-
based
classifier
class,
covering
layout
membership
capabilities,
calculation
rule
strengths,
class
collection.
Finally,
an
instance
real-global
usage
is
provided,
demonstrating
capability
this
method
analyzing
classifying
facts.
The
concludes
with
some
observations
future
ability
making
use
