STID icon indicating copy to clipboard operation
STID copied to clipboard

Code for our CIKM'22 paper Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting.

Results 12 STID issues
Sort by recently updated
recently updated
newest added

Great project! However, when I try to get a prediction with shape `[B, L, N, C]` through STID, I can only get shape `[B, L, N, 1]`. During the time...

node_emb仅仅对随机生成的矩阵进行交互,但是不直接对his_data进行embed。虽然从知乎上看到这个是一直随机embed方式,但是还是从直觉上很难接受,his_data的空间效应没有直接被编码。如果采用nn.Embedding对空间信息同时进行embed效果会怎么样呢,请问这个方面有对比过吗?或者说,直接embed空间效应有什么编程上的困难吗?

from basicts import launch_training 首先,上述会报错。basicts内部似乎没有launch_training 即使改成from basicts.launcher import launch_training,也无法运行额。 更为重要的是,不知道是如何落实到训练和预测的。感觉缺失了一部分,只有数据预处理和模型骨架和参数设置的部分。可能是我没有找到,还望指点。

PEMS 系列数据集包括:PEMS03、PEMS04、PEMS07、PEMS08。为什么没有用PEMS03呢?谢谢。

请问作者代码中 self.scaler = load_pkl("{0}/scaler_in{1}_out{2}.pkl".format(cfg["TRAIN"]["DATA"]["DIR"], cfg["DATASET_INPUT_LEN"], cfg["DATASET_OUTPUT_LEN"])) 使用了这个文件在哪里可以下载

I noticed that only the last time step of each sample is kept by the following code: https://github.com/zezhishao/STID/blob/f9801a5d4939e7d3eb100525c1c027cbd34630d3/stid/stid_arch/stid_arch.py#L76-L88 More specifically, the operation "t_i_d_data[:, -1, :]" is used to get the...

![image](https://github.com/zezhishao/STID/assets/131830057/f55feeda-71ef-4975-8064-ae85e92475ed) ![image](https://github.com/zezhishao/STID/assets/131830057/23876380-4031-49fb-8fd9-193063af7c0a) 为什么安装easy-torch=1.2.10后,安装的pytorch1.10.0被卸载了。

配置文件SEED默认为1,但每次运行的结果都不同是因为什么呢