Hey HN folks -- I am a graduate student exploring the space of time-series forecasting. I wrote a Colab notebook to highlight the challenges in forecasting: what makes it difficult, and how standardized forecasting accuracy metrics can be misleading.
I used the newly introduced EvaDB's forecasting methods (https://evadb.readthedocs.io/en/stable/source/reference/ai/m...) and took on an SQL-like approach to perform forecasting.
Would love to hear your thoughts :) Thanks for your time!
Link to the notebook: https://colab.research.google.com/drive/1Guc04LGZ9ugdysucgVC...