STFGNN icon indicating copy to clipboard operation
STFGNN copied to clipboard

Code of STFGNN@AAAI-2021 (Spatial-Temporal/ Traffic data forecasting)

Results 13 STFGNN issues
Sort by recently updated
recently updated
newest added

请问你是如何复现出Graph wavenet的结果,我自己跑出Graph wavenet的结果比你论文中展示的小很多。

Hi, Thanks for your wonderful works~ Recently, I found there seems to be some inconsistency between [STFGNN](https://github.com/MengzhangLI/STFGNN/blob/a02feee56fbb04d6cabb8f04f625948072fa9583/main_4n0_3layer_12T_res.py#L143) and other baselines (e.g., Graph WaveNet, DCRNN) in the way of calculating metrics,...

Excuse me, could you please talk about the concatenation of the three adjacency matrices?

![捕获](https://user-images.githubusercontent.com/65726684/137275222-56560e38-ee64-477b-839a-c5f99e1c0551.PNG) 您好,请问您可以帮我解释一下这些参数分别代表什么嘛?还有程序的运行流程是怎么样呢?需要我自己在单独的去生成每个数据集时态图吗?

I am trying to use Temporal_Graph_gen.py to process METR-LA. ![6357191CBE7F0D7BFD52581B95012938](https://user-images.githubusercontent.com/65991287/134805723-cf488d9a-396d-4ad1-86ac-3113e21de21a.png) README said that in PEMS08 dataset, the shape is (Total_Time_Steps, Node_Number), however, when I load the PEMS08, the shape is...

您好,由于您宝贵的框架基于STSGCN,所以我想问下关于STSGCN的复现问题。因为现在貌似大家都无法复现它,所以我想问下当时您的复现情况,是不是我们忽略了它的一些细节。若您能提供一些方向也可以,因为那个框架的复现结果和论文结果差的有点远,若您能提供您复现的方向把差距缩小到误差范围内,我们将感激不尽,谢谢。

# 作者,你好。我在PEMS03数据集上运行你们提供的demo,为什么运行出的结果在前2轮看起来很正常,到第3轮的train loss直接变成200多了,直到训练结束也还是200多,没有得到正确的实验结果。dtw文件就是adj_PEMS03_001.csv吧,我也尝试用fast_DTW_gen.py重新生成adj_PEMS03_001.csv,但还是在第3轮遇到同样的问题,请问原因是什么?

![L%H56PL`R734NWH32H9M8JW](https://user-images.githubusercontent.com/71821906/185900994-48b441ec-8303-4bd3-ad69-d4a87ea77904.png) 想请问一下这个公式里各个元素的shape,不能理解。在STSGCN中有个一模一样的公式,但是那个shape很清楚。主要就是h的shape是多少不太明白。感谢