Advantag Disadvant Future

SNo Title Method Result

Improve
Struggles
High spatial
xTP-LLM Achieved with
accuracy, modeling,
Towards (based on MAE of highly
explainabl integrate
Explainabl Llama2- 21.91 and dynamic
e outputs, real-time
e Traffic 7B-chat, MAPE of traffic
1 robust incident
Flow fine- 11.01, anomalies
across data,
Prediction tuned outperfor ; requires
temporal/ expand to
with LLMs with ming 9 large
spatial broader
LoRA) baselines textual
domains urban
prompts
analytics

Integratio
n of deep
Simple
Effective Limited to learning,
Regressio implemen
Traffic hourly regressio neural
n tation,
Prediction traffic n; not networks,
(Sklearn, real-time
2 using prediction robust for big data;
Keras, prediction
Machine ; accuracy dynamic real-time
TensorFlo , uses
Learning evaluated or sparse GPS and
w) open
using MSE data weather
datasets
data
analysis

Developm
ent of
Wide Inconsiste hybrid,
Artificial SVM, coverage nt dataset real-time,
Accuracy
intelligen KNN, RF, of ML/DL availabilit and big-
up to
ce-based CNN, models, y, limited data
98.6%
3 traffic LSTM, applicable model integrate
(model-
flow RNN, Q- to ITS, generaliza d models;
dependen
prediction Learning good tion, high better
t)
: a review etc. prediction resource spatial-
accuracy demand temporal
adaptatio
n
 Apply
model in
ARIMA, real-
Integrates No
KNN, world
Enhancin Theoretic historical, empirical
Deep urban
g Traffic al model; real-time, testing;
Neural networks,
Prediction 85% and ETA theoretic
Networks test
4 with accuracy data; al only;
(DNN), scalability
Historical in adaptable needs
ML-based ,
Data and reference and multi- real-
ETA integrate
ETA d work model world
estimatio live
approach validation
n datasets
for
validation

Empirical
Integrates
testing,
multiple
Enhancin ARIMA, Theoretic real-time
models Lacks
g Traffic KNN, al model; integratio
and data empirical
Prediction DNN, ML- ~85% n, and
types; validation
5s with based accuracy deployme
adaptable ;
Historical ETA in nt in real-
, forward- theoretic
Data and estimatio reference world
looking al only
ETA n d studies urban
traffic
traffic
insights
systems