DePaul's CSC 578 (Neural Networks) course Final Project's Kaggle Competition www.kaggle.com/competitions/csc-578-final-project-spring-2024/overview/description
The goal of the project is to apply deep learning to do time series forecasting. In particular, you create deep learning models to predict future traffic volume at a location in Minnesota, between Minneapolis and St Paul. The dataset is a cleaned-up version of UCI for hosting the Metro Interstate traffic volume.
The specific goal will be to predict from a 12-hour input window, just the traffic volume for 3 hours past the end of the window. That is a little different from the single-step, multi-step, and multi-output RNN with LSTM examples from the TensorFlow Time Series Tutorial, but it will be much simpler to compare results within Kaggle. For further details, see the detailed task description.
The evaluation metric for this competition is Mean Absolute Error (MAE).
For every author in the dataset, submission files should contain two columns: id and prediction, where id should be 0-based indices for the rows in the test set, and prediction should be the predicted traffic volume of the row. The id number/index should start with 35589. The submission file must start with a header row, and be followed by 4986 rows of predictions (for ids from 35589 to 40574). No comma within id numbers predictions should be used -- commas should be used only as a delimiter between the two columns. Also there should be no space between the two columns..
id,prediction 35589,3481.280499629291 35590,2876.958684404887 35591,3417.846569017333 35592,3293.5700851714946 ... 40572,3947.3713909650096 40573,3542.831172174818 40574,2781.3480226893225