wenmellors / ts-trajgen Goto Github PK

Hi, thank you for sharing your codes! Could you please share how to get or generate od_distinct_route.json files? They are used in train_gan.py and train_region_gan.py

Can you kindly provide this file? 'tmppartition50' . It is used in "construct_traffic_zone_relation.py", but is missed now.
Also, the class and function "tarjan.tarjan" is missed.
Why don't you provide sample data in the code hub? It would be much easier for us to transfer this model to our work.

Hi, thanks for sharing your codes! Could you please share how to calculate the metrics (The metric definition is in the appendix but the available paper lacks of the appendix) ? Or could you please offer the evaluation.py?

Hello @WenMellors, Thank you for the detailed guidance.
I am following your instructions to run the model based on BJ_Taxi dataset. But when I ran the construct_region_dist.py, there was one file named "xianshi_partA_traj_mm_processed.csv", but did not show where it comes from. I could not use it as a reference to deal with the BJ_Taxi dataset.
May I have your help to explain where this CSV file is from please? Thank you so much!

您好！
共有两个问题希望您能解答，
1、train_gan.py 中使用的真实轨迹数据集是201511_week1_short_traj，而 train_region_gan.py中使用的真实轨迹数据集是201511_week1_mm_region_test.csv ，请问这个真实轨迹数据集是用训练数据集，测试数据集还是未划分（包含所有轨迹数据）的数据集。
2、训练中训练判别器很快，使用1个3080训练生成器（生成2000条轨迹）需要13个小时，这是否正常呢？
若能解答，不胜感激。

I appreciate your paper and the clean style codes, but I haven't find any dataset on this github repository. Could you please put them on the repository too if convenient ? Thanks a lot !

Hello @WenMellors, thank you and your colleagues for your efforts in creating this well-crafted repository. While reviewing the code, I noticed there are a few datasets, including some .json files, that are currently inaccessible. I understand the potential privacy considerations associated with this. However, this has made it a bit challenging for me to replicate the project completely.

Would it be possible to provide a few sample data or an outline of the data structure? This would greatly assist in understanding how to organize our own data effectively. Sharing such insights would undoubtedly benefit the community and enhance the impact of your work.

Thank you in advance for your attention.

Thank you for opening and maintaining this project.

I want to reference your paper and reproduce your experimental results. However, I find that the region-level files for pretraining (eg. 201511_region_pretrain_input_train.csv) cannot be found. May I ask if they are consistent with the content of road-level files for pretraining (eg. bj_taxi_pretrain_input_train.csv), or are they processed data obtained through some type of mapping?

Thank you very much for your answer. I will continue to pay attention to your work!

我在尝试使用你们的项目的时候遇到了许多困难，比如运行preprocess_pretrain_input.py时，发现你们没有划分训练集、测试集，我擅自划分了下；然后在下一步运行pretrain_gat_fc.py时发现，你们没有adjacent_mx.npz和node_feature.pt等文件，请问这些数据在哪里可以获取？如果是你们自己处理的数据，请问方便提供一下吗？

作者您好，我是武汉大学2023级的一名研究生。

最近看到您团队所发表的NASR这篇文章，《Empowering A* Search Algorithms with Neural Networks for Personalized Route Recommendation》。您的团队在github上开源了代码，但没有放完整的数据集，代码里的所需的输入数据和所公开的云盘里的数据不一致，因此导致代码无法看懂与运行（例如trainData、trainGeoData、trainTimeData这些变量含义未知）。

我曾向几位作者发过邮件，但都没有回复。请问您可以帮忙联系下他们吗？

这篇文章我们想用来做基线，因为路径规划开源的代码非常少，所以它对我们很重要，希望您能回复我们。

祝您龙年大吉，工作愉快，生活如意，日日顺心~~~

Hello @WenMellors, first of all I'd like to thank you (and your colleagues) for the hard work put into this library.
In the codes of the library, I found that the majority part is for the Xian's dataset to generate different prerequired files. May I ask if I can have the access to the Xian's datset by any chance? I checked the LibCity library and it showed that "Please contact the author" for the Xian trajectory dataset.

I see you use adjacent_list.json in the file preprocess_pretrain_input.py, but I cannot find the file, can you provide that