bug: Flowstate cannot forecast when the condition is h=1

[JitaekJo]

Initial Checklist

• I confirm that I'm using the latest version of TimeCopilot.
• I confirm that I've reviewed my issue at https://github.com/AzulGarza/timecopilot/issues before opening this one.

What happened?

Hi,

please see the below error . I ran it locally.
When the h=2, the model returns correctly.

--> 102 res = tcf.forecast(
103 df=df,
104 h=1,
105 freq="W"
106 )
108 def explain_with_llm(table_md: str, prompt=
109 """
110 예측 결과를 표로 정리해줘.
(...)
113
114 """):
115 base = os.environ.get("OPENAI_BASE_URL", "http://xxx.xxx.xx.xxx:11434/v1").rstrip("/")

File c:\Users\jitaek.jo\AppData\Local\Programs\Python\Python312\Lib\site-packages\timecopilot\forecaster.py:177, in TimeCopilotForecaster.forecast(self, df, h, freq, level, quantiles)
122 def forecast(
...
--> 196 fcsts_mean_np = np.concatenate(fcsts_mean_tp)
197 if quantiles is not None:
198 fcsts_quantiles_np = np.concatenate(fcsts_quantiles_tp)

ValueError: zero-dimensional arrays cannot be concatenated

Relevant log output

Environment

0.0.20

[Kushagra7777]

hey @JitaekJo
Thank you so much for using TimeCopilot.
We tested out flowstate forecast for h=1.
Take a look at this notebook
specifically last cell, the forecast works well for h=1
It's could be possible that there might be some issue with the dataset you are feeding to TimeCopilot forecaster after pre-processing. --> Could you please share the dataset with us, if not can you share other details like: structure, data type and all.
--> If possible, please share the pipeline you are using.

[Jigit-ship-it]

Brent_v3(M).xlsx

Hi here is the file which I feed
Please be noted that I ran Flowstate locally with Ollama

[Kushagra7777]

Hey @Jigit-ship-it
Install this branch instead of main one https://github.com/AzulGarza/timecopilot/tree/issue_242
We have resolved the bug here. And do let us know your feedback.

thanks

[Jigit-ship-it]

Heyy

Thank you so much, it works!