Open uliuycy opened 1 month ago
can you share your energy dataset?
It's the dataset in your code
https://github.com/abudesai/timeVAE/blob/main/datasets/energy_subsampled_train_perc_10.npz Is there a problem with my T-SNE code. I am using the visualization code from Timegan, and the original data is scaled trained data. The generated data is the data generated after normal distribution sampling.
Hello, is it possible to compare the data generated after sampling the heavy normal distribution with the training data in a t-sne plot
I have updated the repo. Sorry, it's a major overhaul because we are now planning to work on the next version of TimeVAE so we wanted to clean things up. If you run src/vae_pipeline.py
, it will run through all the datasets, train the model, generate samples, and create and save the t-sne plots for each dataset.
Why is the graph I generated from energy data running like this ![image](https://github.com/abudesai/timeVAE/assets/149740828/29fec225-29a2-4701-bafd-60443601ff2d ![屏幕截图 2024-05-17 092813](https:
//github.com/abudesai/timeVAE/assets/149740828/48ed3a52-eca8-48f8-88b8-e642676de0e5)
ori_data = np.asarray(X)
generated_data = np.asarray(samples)