Hybrid Deep Learning Models for Predicting Student Academic Performance
Mathematical and Computational Applications,
Год журнала:
2025,
Номер
30(3), С. 59 - 59
Опубликована: Май 23, 2025
Educational
data
mining
(EDM)
is
instrumental
in
the
early
detection
of
students
at
risk
academic
underperformance,
enabling
timely
and
targeted
interventions.
Given
that
many
undergraduate
face
challenges
leading
to
high
failure
dropout
rates,
utilizing
EDM
analyze
student
becomes
crucial.
By
predicting
success
identifying
at-risk
individuals,
provides
a
data-driven
approach
enhance
performance.
However,
accurately
performance
challenging,
as
it
depends
on
multiple
factors,
including
history,
behavioral
patterns,
health-related
metrics.
This
study
aims
bridge
this
gap
by
proposing
deep
learning
model
predict
with
greater
accuracy.
The
combines
convolutional
neural
network
(CNN)
bidirectional
gated
recurrent
unit
(BiGRU)
predictive
capabilities.
To
improve
model’s
performance,
we
address
key
preprocessing
challenges,
handling
missing
data,
addressing
class
imbalance,
selecting
relevant
features.
Additionally,
incorporate
optimization
techniques
fine-tune
hyperparameters
determine
best
architecture.
Using
metrics
such
accuracy,
precision,
recall,
F-score,
our
experimental
results
show
proposed
achieves
improved
prediction
accuracy
97.48%,
90.90%,
95.97%
across
three
datasets.
Язык: Английский
A high ranking-based ensemble network for student’s performance prediction using improved meta-heuristic-aided feature selection and adaptive GAN for recommender system
Kybernetes,
Год журнала:
2024,
Номер
unknown
Опубликована: Авг. 24, 2024
Purpose
Predicting
student
performance
is
crucial
in
educational
settings
to
identify
and
support
students
who
may
need
additional
help
or
resources.
Understanding
predicting
essential
for
educators
provide
targeted
guidance
students.
By
analyzing
various
factors
like
attendance,
study
habits,
grades,
participation,
teachers
can
gain
insights
into
each
student’s
academic
progress.
This
information
helps
them
tailor
their
teaching
methods
meet
the
individual
needs
of
students,
ensuring
a
more
personalized
effective
learning
experience.
identifying
patterns
trends
performance,
intervene
early
address
any
challenges
acrhieve
full
potential.
However,
complexity
human
behavior
makes
it
difficult
accurately
forecast
how
will
perform.
Additionally,
availability
quality
data
vary,
impacting
accuracy
predictions.
Despite
these
obstacles,
continuous
improvement
collection
development
robust
predictive
models
enhance
effectiveness
scalability
existing
different
populations
be
hurdle.
Ensuring
that
are
adaptable
across
diverse
environments
widespread
use
impact.
To
implement
performance-based
recommendation
scheme
capabilities
suggesting
better
materials
papers,
books,
videos,
hyperlinks
according
needs.
It
enhances
higher
education.
Design/methodology/approach
Thus,
approach
achievement
presented
using
deep
learning.
At
beginning,
accumulated
from
standard
database.
Next,
collected
undergoes
stage
where
features
carefully
selected
Modified
Red
Deer
Algorithm
(MRDA).
After
that,
given
Deep
Ensemble
Networks
(DEnsNet),
which
techniques
such
as
Gated
Recurrent
Unit
(GRU),
Conditional
Random
Field
(DCRF),
Residual
Long
Short-Term
Memory
(Res-LSTM)
utilized
performance.
In
this
case,
parameters
within
DEnsNet
network
finely
tuned
by
MRDA
algorithm.
Finally,
results
obtained
superior
method
delivers
final
prediction
outcome.
Following
Adaptive
Generative
Adversarial
Network
(AGAN)
introduced
recommender
systems,
with
optimally
Lastly,
evaluated
numerically
compared
traditional
demonstrate
proposed
approach.
Findings
The
developed
model
7.66%,
9.91%,
5.3%,
3.53%
than
HHO-DEnsNet,
ROA-DEnsNet,
GTO-DEnsNet,
AOA-DEnsNet
dataset-1,
7.18%,
7.54%,
5.43%
3%
enhanced
dataset-2.
Originality/value
recommends
appropriate
short
period
improve
ability.
Язык: Английский
Agricultural price prediction based on data mining and attention-based gated recurrent unit: a case study on China’s hog
Journal of Intelligent & Fuzzy Systems,
Год журнала:
2024,
Номер
46(4), С. 9923 - 9943
Опубликована: Март 26, 2024
Under
the
influence
of
coronavirus
disease
and
other
factors,
agricultural
product
prices
show
non-stationary
non-linear
characteristics,
making
it
increasingly
difficult
to
forecast
accurately.
This
paper
proposes
an
innovative
combinatorial
model
for
Chinese
hog
price
forecasting.
First,
is
decomposed
using
Seasonal
Trend
decomposition
Loess
(STL)
model.
Next,
data
are
trained
with
Long
Short-term
Memory
(LSTM)
Autoregressive
Integrated
Moving
Average
(SARIMA)
models.
Finally,
prepared
multivariate
factors
after
Factor
analysis
predicted
gated
recurrent
neural
network
attention
mechanisms
(AttGRU)
obtain
final
prediction
values.
Compared
models,
STL-FA-AttGRU
produced
lowest
errors
achieved
more
accurate
forecasts
prices.
Therefore,
proposed
in
this
has
potential
forecasting,
contributing
development
precision
sustainable
agriculture.
Язык: Английский
Predictive Models for Educational Purposes: A Systematic Review
Big Data and Cognitive Computing,
Год журнала:
2024,
Номер
8(12), С. 187 - 187
Опубликована: Дек. 13, 2024
This
systematic
literature
review
evaluates
predictive
models
in
education,
focusing
on
their
role
forecasting
student
performance,
identifying
at-risk
students,
and
personalising
learning
experiences.
The
compares
the
effectiveness
of
machine
(ML)
algorithms
such
as
Support
Vector
Machines
(SVMs),
Artificial
Neural
Networks
(ANNs),
Decision
Trees
with
traditional
statistical
models,
assessing
ability
to
manage
complex
educational
data
improve
decision-making.
search,
conducted
across
databases
including
ScienceDirect,
IEEE
Xplore,
ACM
Digital
Library,
Google
Scholar,
yielded
400
records.
After
screening
removing
duplicates,
124
studies
were
included
final
review.
findings
show
that
ML
consistently
outperform
due
capacity
handle
large,
non-linear
datasets
continuously
enhance
accuracy
new
patterns
emerge.
These
effectively
incorporate
socio-economic,
demographic,
academic
data,
making
them
valuable
tools
for
improving
retention
performance.
However,
also
identifies
key
challenges,
risk
perpetuating
biases
present
historical
issues
transparency,
complexity
interpreting
AI-driven
decisions.
In
addition,
reliance
varying
processing
methods
reduces
generalisability
current
models.
Future
research
should
focus
developing
more
transparent,
interpretable,
equitable
while
standardising
collection
incorporating
non-traditional
variables,
cognitive
motivational
factors.
Ensuring
transparency
ethical
standards
handling
is
essential
fostering
trust
Язык: Английский
Exploring the Impact of Elevated Learning Methodology on Student Performance Prediction: An Empirical Analysis
L. Priyadharshini,
K. Niranjana
Опубликована: Апрель 18, 2024
Язык: Английский
Enhancing tertiary students’ programming skills with an explainable Educational Data Mining approach
PLoS ONE,
Год журнала:
2024,
Номер
19(9), С. e0307536 - e0307536
Опубликована: Сен. 3, 2024
Educational
Data
Mining
(EDM)
holds
promise
in
uncovering
insights
from
educational
data
to
predict
and
enhance
students’
performance.
This
paper
presents
an
advanced
EDM
system
tailored
for
classifying
improving
tertiary
programming
skills.
Our
approach
emphasizes
effective
feature
engineering,
appropriate
classification
techniques,
the
integration
of
Explainable
Artificial
Intelligence
(XAI)
elucidate
model
decisions.
Through
rigorous
experimentation,
including
ablation
study
evaluation
six
machine
learning
algorithms,
we
introduce
a
novel
ensemble
method,
Stacking-SRDA,
which
outperforms
others
accuracy,
precision,
recall,
f1-score,
ROC
curve,
McNemar
test.
Leveraging
XAI
tools,
provide
into
interpretability.
Additionally,
propose
identifying
skill
gaps
among
weaker
students,
offering
recommendations
enhancement.
Язык: Английский