Traffic prediction using time series models is important for forecasting the volume and density of traffic flow, usually for the purpose of managing vehicle movement, reducing congestion, and generating the optimal route. This work will analyze the performances of different predictive time-series models (ARIMA, SARIMA, LSTM) for predicting traffic flow. The performance of each model is evaluated based on the error functions incurred by each approach. It is necessary to understand how to choose the right combination algorithms and the dataset approachable. So, in this research we are going to explore how to analyze time-series models and algorithms and how machine learning helps to forecast congestion and plan optimal routes. This work attempts to highlight the usefulness of Time Series analysis in traffic forecasting by using multivariate and univariate analysis to understand the structure of the data to choose the right modeling technique. The result showed which technique is the best to model the time-series traffic conditions

Predictive Modeling; Time-series Analysis; Traffic Flow; LSTM; ARIMA; SARIMA