Evaluating Clustering Algorithms: An Analysis using the EDAS Method
E3S Web of Conferences,
Год журнала:
2023,
Номер
430, С. 01161 - 01161
Опубликована: Янв. 1, 2023
Data
clustering
is
frequently
utilized
in
the
early
stages
of
analyzing
big
data.
It
enables
examination
massive
datasets
encompassing
diverse
types
data,
with
aim
revealing
undiscovered
correlations,
concealed
patterns,
and
other
valuable
information
that
can
be
leveraged.
The
assessment
algorithms
designed
for
handling
large-scale
data
poses
a
significant
research
challenge
across
various
fields.
Evaluating
performance
different
processing
yield
or
even
contradictory
results,
phenomenon
remains
insufficiently
explored.
This
paper
seeks
to
address
this
issue
by
proposing
solution
framework
evaluating
algorithms,
objective
reconciling
divergent
conflicting
evaluation
outcomes.
“The
multicriteria
decision
making
(MCDM)
method”
used
assess
algorithms.
Using
EDAS
rating
system,
report
examines
six
alternative
“the
KM
algorithm,
EM
filtered
(FC),
farthest-first
(FF)
make
density-based
(MD),
hierarchical
(HC)”—against,
external
measures.
Expectation
Maximization
(EM)
algorithm
has
an
ASi
value
0.048021
ranked
5th
among
Farthest-First
Algorithm
0.753745
2nd.
Filtered
Clustering
(FC)
0.055173
4th.
Hierarchical
(HC)
highest
0.929506
1st.
Make
Density-Based
(MD)
0.011219
6th.
Lastly,
K-Means
0.055376
3rd.
These
values
provide
each
algorithm’s
overall
performance,
rankings
offer
comparative
analysis
their
performance.
Based
on
result,
we
observe
achieves
first,
indicating
its
superior
compared
Язык: Английский
A New Method Using Artificial Neural Networks to Group Mines into Similar Sets for Efficient Management and Transformation
Applied Sciences,
Год журнала:
2024,
Номер
14(8), С. 3350 - 3350
Опубликована: Апрель 16, 2024
The
market
economy
means
that
only
those
companies
are
characterised
by
the
generation
of
positive
economic
results
and
liquidity
can
function,
survive
thrive.
Due
to
importance
coal
industry
in
social
terms—due
number
people
employed
industry—it
is
necessary
constantly
search
for
methods
improve
management
business
efficiency.
This
paper
proposes
use
artificial
neural
networks
group
mines
into
sets
similar
mines.
These
be
used
make
different
decisions
these
companies.
sites
easily
compared
with
each
other,
areas
need
restructured.
In
addition,
developing
pro-efficiency
strategies
designated
groups
simpler
than
mine
individually.
reduces
such
studies
real
terms
allows
effective
measures
applied
more
quickly.
Язык: Английский
Robust Semi-Parametric Inference for Two-Stage Production Models: A Beta Regression Approach
Symmetry,
Год журнала:
2023,
Номер
15(7), С. 1362 - 1362
Опубликована: Июль 4, 2023
The
data
envelopment
analysis
is
related
to
a
non-parametric
mathematical
tool
used
assess
the
relative
efficiency
of
productive
units.
In
different
studies
on
efficiency,
it
common
employ
semi-parametric
procedures
in
two
stages
determine
whether
any
exogenous
factors
interest
affect
performance
However,
some
these
procedures,
particularly
those
based
conventional
statistical
inference,
generate
inconsistent
estimates
when
dealing
with
incoherent
data-generating
processes.
This
inconsistency
arises
due
scores
being
limited
unit
interval,
and
estimated
often
exhibit
serial
correlation
have
observations.
To
address
such
inconsistency,
several
strategies
been
suggested,
most
well-known
an
algorithm
parametric
bootstrap
procedure
using
truncated
normal
distribution
its
regression
model.
this
work,
we
present
modification
that
utilizes
beta
structure.
model
allows
for
better
accommodation
asymmetry
distribution.
Our
proposed
introduces
inferential
characteristics
are
superior
original
algorithm,
resulting
more
statistically
coherent
process
improving
consistency
property.
We
conducted
computational
experiments
demonstrate
improved
results
achieved
by
our
proposal.
Язык: Английский