search

Nixtla / nixtla

TimeGPT-1: production ready pre-trained Time Series Foundation Model for forecasting and anomaly detection. Generative pretrained transformer for time series trained on over 100B data points. It's capable of accurately predicting various domains such as retail, electricity, finance, and IoT with just a few lines of code 🚀.
https://docs.nixtla.io
Other
1.91k stars 151 forks source link

pandas incompatibility #193

Closed davecmm closed 6 months ago

davecmm commented 6 months ago

Looks good so far!, no problem with forecasting.

Following along this quick start guide tho', and when using timegpt.plot:

timegpt.plot(df, timegpt_fcst_df, time_col='timestamp', target_col='value')

throws error:

~/.local/lib/python3.10/site-packages/pandas/core/indexers/utils.py in disallow_ndim_indexing(result) 341 """ 342 if np.ndim(result) > 1: --> 343 raise ValueError( 344 "Multi-dimensional indexing (e.g. obj[:, None]) is no longer " 345 "supported. Convert to a numpy array before indexing instead."

ValueError: Multi-dimensional indexing (e.g. obj[:, None]) is no longer supported. Convert to a numpy array before indexing instead.

Fallen back on good ole fashioned matplotlib for now ....

jmoralez commented 6 months ago

Hey. This is due to an incompatibility between your installed versions of pandas and matplotlib, not a problem with the library. You should be able to solve it by upgrading matplotlib.

davecmm commented 6 months ago

doh! I'm an idiot, thanks.

jmoralez commented 6 months ago

I don't think you were completely at fault here, we only had matplotlib as the requirement, we've changed it to be pandas[plot] instead. That should check the compatibility when installing, hopefully avoiding this case. Thanks for reporting!