Intelligent Traffic Forecast (Ali Tianchi)
Top 3%
Ali Tianchi Intelligent Traffic Forecast Challenge
Challenge:
Under the theme of ‘Smart Traffic Prediction in the Era of Mobile Internet,’ participants are required to develop algorithm models based on internet traffic information to accurately predict the average travel time of vehicles on key road segments during specific time periods. The goal is to produce travel time prediction algorithms that leverage big data techniques to mitigate the impact of traffic congestion on urban management. This will better assist Guiyang’s traffic managers in formulating control strategies in advance, preventing and reducing congestion, and achieving smart travel solutions.
Solution:
- Developed a time sequence model to estimate the average travel time between 7 AM to 8 AM in July, utilizing historical data of daily travel times for each vehicle across 132 roads from March to May.
- Utilized LSTM for prediction, meticulously refining parameters such as encoder and decoder layers, hidden units, batch size, and dropout for optimal model performance.