search

AI4HealthUOL / SSSD

Repository for the paper: 'Diffusion-based Time Series Imputation and Forecasting with Structured State Space Models'
MIT License
273 stars 52 forks source link

How to draw plots_merged with code? #17

Closed YZYZYZY closed 1 year ago

YZYZYZY commented 1 year ago

Thank you for your wonderful work！ I want to use the results I ran out to draw the sequence comparison diagram in your paper, but my code ability is limited and it is not as beautiful as your drawing anyway. Could you please provide the code for drawing the diagram? Thank you so much!

juanlopezcode commented 1 year ago

thank you @YZYZYZY, I don't get it, do you want to draw an architecture diagram? I made mine in https://app.diagrams.net/, or do you want o actually made an extension of our current architecture? (thus you want the existing one to make minimal changes)?

YZYZYZY commented 1 year ago

I want to learn how to draw such a result sequence comparison diagram for visual comparison of the results, but I can't find the corresponding code in the code you provided. I'd be grateful if you could provide that. image

juanlopezcode commented 1 year ago

unfortunatelly we didn't release the plotting procedure, but the signals are 4: -plot original signal 'target' -plot the median imputation of n samples mutiplied by nans where missing (mask it) as 'ssssd' -plot (fill_between) the 0.25 and 0.75 quantiles over the n samples again multiplied by nans (masked) with dark colour (green here) -plot the 0.025 and 0.975 quantiles masked again with light colour (green here) -additionally plot (axvspan) the background shadded area where missing with ligh colour

hope it helps