search

AutoViML / Auto_TS

Automatically build ARIMA, SARIMAX, VAR, FB Prophet and XGBoost Models on Time Series data sets with a Single Line of Code. Created by Ram Seshadri. Collaborators welcome.
Apache License 2.0
723 stars 113 forks source link

Testdata error #69

Closed Hydaspex closed 2 years ago

Hydaspex commented 2 years ago

Hi gents, When i run model. Predict a silly error comes up. It is related to the parameter testdata and returns object of type float has no len() Please let me know if you have a quick solution. Thanks for your help.

AutoViML commented 2 years ago

Hi @Hydaspex 👍 It is fixed now. Please do the upgrade to latest version by: pip install auto-ts --upgrade

Please confirm that fixes the problem. Thanks Auto_ViML

Hydaspex commented 2 years ago

Yes it works now however there seems to be an issue in processing the testdata's index when the model is ML. The error message is quoted below:

redicting using test dataframe as input for ML model For large datasets: ML predictions will take time since it has to predict each row and use that for future predictions... Using given input: pandas dataframe... converting testdata to datetime index erroring. Please check input and try again. ML predictions completed

Thanks

AutoViML commented 2 years ago

Hi @Hydaspex 👍 Can you please post a Colab notebook or GitHub link for your notebook where you re getting this error. It would help me find out the issue quickly. Can you please give me more input here?

Hydaspex commented 2 years ago

Hi @AutoViML,

you can find the notebook at the following repository

https://github.com/Hydaspex/rrt_waiting_times.git

Thanks

rsesha commented 2 years ago

Hi @Hydaspex 👍 I noticed that you have set the time index already. Instead, I commented it out. Just send the dataframe as is and auto-ts will automatically set the time index. Once you make that one line change, it works automatically on train and test. Take a look at the notebook below. rtt_model_v2.zip

One more thing: upgrade to the latest version and test it: pip install auto-ts --upgrade

Close this issue if it is fixed. AutoViML team

geRupak commented 2 years ago

For large datasets: ML predictions will take time since it has to predict each row and use that for future predictions... Model is erroring during prediction. Check inputs and try again. how can i slove this error.