MSBD5002 Highway Tollgates Traffic Flow Prediction Course Project

Introduction

1. The programming language used in this project is python 3.
2. The packages used in this project including: pandas, numpy, sklearn, scipy, matplotlib, seaborn, pytorch, skorch, xgboost, lightgbm, etc.

Project Details

Data processing

Codes about Loading data and Data preprocessing, original data and processed data are kept in dataprocessing folder.

1. Run Previous 2-hour train.ipynb, it can generate '2h_route_avg_train.xlsx'.
2. Run Previous 2-hour test.ipynb, it can generate 2h_route_avg_test.xlsx.
3. Run 5002_data_preprocessing.ipynb, it can generate training_nn.csv and test_data.csv.
4. These files already exist in processed_data folder, to rewrite these files, you need to uncomment the corresponding code block.

Models and evaluation

Codes about Prediction algoritm and Performance and our models are kept in models folder.

1. Run Training part in Train and Ensemble Models.ipynb, it can generate .pkl files of our models.
2. Run Ensemble model and Evaluation in validation dataset part in Train and Ensemble Models.ipynb, it can load pre-trained models and get responding results and perform prediction on validation data and compute MAPE of the ensemble model.
3. Run Prediction on test data part in Train and Ensemble Models.ipynb, it can generate submission.csv.

Submission

Our submission file is kept in submission folder named submission.csv.