Raw Traffic Sensor Data

(PeMS, METR-LA)

Data Cleaning

(Missing Value Handling, Denoising)

Normalization & Feature Scaling

(MinMax / Z-Score)

Sliding Window Generator

(Timesteps for Prediction)

Sequential Input

(Past T Timesteps)

Contextual Features

(Time of Day, Day of Week)

Feature Concatenation

(GRU Output + Context)

GRU Network

Temporal Feature Learning

GRU Output Vector

(Encoded Features)

Tree-Based Model

XGBoost Regressor

Predicted Traffic Flow

(Next 15/30/60 Min)