search

time-series-foundation-models / lag-llama

Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting
Apache License 2.0
1.09k stars 121 forks source link

Bad accuracy with the wide dataframe #17

Closed TLF-1234 closed 2 months ago

TLF-1234 commented 4 months ago

I am having bad accuracy (MSE of 2500+)

image

I am giving the model 10 days worth of data and using the default settings given in the updated colab with a wide dataframe. Prediction length = 24

While i see improvement if i increase the prediction length i.e. inplace of predicting for 1 day i predict for 72 time stamps (3 days) that i can get a mse of 400. Which is still high and not as good.

This image is 3 days prediction: image

Do let me know if i am making a large error when giving this data which is causing this low accuracy.

ashok-arjun commented 4 months ago

Can you please share a demo notebook with the dataset? You can possibly use a dummy dataset in the same style if you want to keep your data private.

It is also possible that the zero-shot forecasting performance is bad on this dataset. I'd suggest to wait for the finetuning code and try it out then.

clanert27 commented 4 months ago

Can you please share a demo notebook with the dataset? You can possibly use a dummy dataset in the same style if you want to keep your data private.

It is also possible that the zero-shot forecasting performance is bad on this dataset. I'd suggest to wait for the finetuning code and try it out then. @ashok-arjun When can we expect the finetuning code to be released?

ashok-arjun commented 3 months ago

Apologies for the delay @clanert27, you can expect the full release in 2 weeks approximately.

ashok-arjun commented 2 months ago

Hi @clanert27 @TLF-1234 Can you please try finetuning now? We have added a list of instructions and best practices in the README that you could follow.

ashok-arjun commented 2 months ago

Closing this issue as it is resolved. Feel free to open it again if required :)