Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add a model forecasting notebook in tutorials (#1118)
Add a notebook for forecasting. Blocked by - [x] #1117 Bypassed by installing evadb from a commit before the PR. - [x] #1119 Bypassed by using a nested SQL query, will improve later. This PR adds a basic version of notebook for forecasting. We have also found several issues we need to improve in later PRs: **EvaDB related:** 1. Nested SQL query must have AS, which is redundant. 2. Can not order by a column which is the function's output from the projection's target list. 3. When we are tuning the parameters that used for training, we need to run the drop function every time. **Statsforecast integration related:** 1. Math domain error with statsforecast, when we want to forecast property prices for different postcode. 2. Auto inferring frequency never works. 3. Changing the frequency does not seem to affect the output of the forecasting value, which is unexpected. 5. The catalog column can only be one column, so we need to train three different models for postcode, bedrooms, and propertytype. We should be able to train one model for every combination of these columns.
- Loading branch information