TRAFFIC ROUTE PREDICTION

Abstract

We present a Traffic Route Prediction (TRP)
system aimed at predicting alternate routes in
response to traffic congestion using machine
learning algorithms. We employ Support
Vector Machines (SVM), Decision Tree, and
Random Forest to analyze traffic data and
predict optimal routes. The performance of
each algorithm is evaluated using accuracy,
precision, recall, F1 score, and confusion
matrix analysis. Decision Tree outperformed
other models in accuracy and robustness,
while SVM showed suboptimal results. The
proposed system leverages real-time data from
GPS, traffic cameras, and mobile sensors to
ensure accurate and timely route
recommendations. Our results demonstrate
the potential of machine learning-based TRP
systems to improve urban mobility and reduce
travel time. The performance of each
algorithm is meticulously assessed using
metrics like accuracy, precision, recall, F1
score, and visualized through confusion
matrix graphs. Prior to applying machine
learning, a thorough data analysis is
conducted through graph visualization to
comprehend the traffic flow and congestion
across various routes. Despite the
comprehensive analysis, SVM exhibits
suboptimal performance, while Decision Tree
stands out as a robust performer in accurately
predicting routes. The Random Forest
algorithm is also leveraged to enhance
prediction robustness. The evaluation metrics
offer a comprehensive understanding of the
algorithms' strengths and weaknesses. The
dataset utilized for training encompasses
detailed information on traffic congestion,
ensuring the reliability and accuracy of the
route predictions. This project not only
contributes to optimizing passenger travel
experiences but also holds potential for
effective traffic management strategies.