sadvaseb
commented
3 years ago Hi, 1) We do not need surrounding vehicles' information to calibrate an IDM model. We only need basic information. IDM_calibration_all_lane_data1 is the restructured NGSIM with some manual cleanings. 2) I uploaded a new file (Initial GA 1.31 decimal limit cross-validation.py) which calibrates the IDM model.
I have some codes which need lots of cleaning to be uploaded here and be readable by others. I believe my codes on Github are the best I have for now :) I can read more papers by searching on Google Schalor.
Hi sir, thank you for answering my questions in your another repository "Machine-Learning-and-Deep-Learning-for-Car-Following" again. I have got the dataset NGSIM I-80, including the original one and the restructured one. Then I have some new questions: 1、You say this repository is to extend the Reconstructed NGSIM dataset to include surrounding vehicles, but the dataset you use in models is named "IDM_calibration_all_lane_data1.csv". So I don't know waht's the contact between the csv files this repository output with the file "IDM_calibration_all_lane_data1.csv". 2、In your article “Human Car-Following Behavior: Parametric, Machine-Learning, and Deep-Learning Perspectives”, you use 4 car following models(IDM、FNN、RNN、Deep-RNN). I think "ANN 5 layer with acceleration to headway 1.3.1.py" is for FNN, "Single Layer RNN LSTM with accleration to headway 1.3.1.py" is for RNN, and "Deep Learning RNN with LASTM accleration to headway 1.3.1.py" is for Deep-RNN, but what's for IDM?
I'm new for car following domain, so my questions may be immature, hoping u can understand and answer me. Last and sincerely, if u have some other code or learning materials of CF models and articles about CF can be shared, could u please send them to me. I will be very grateful. My email address is jialiding@zjut.edu.cn. You can just call me Ding and Hoping we can communicate by email in the future. Have a good day!