Patricio Cerda-Mardini
Patricio Cerda-Mardini
Indeed, I think the user wanted the integration to be as "native" as possible (so, basically a new handler).
@ZoranPandovski @Ricram2 How should we communicate this feature? A tutorial in the learning hub? Docs?
@ea-rus any ideas? I would suggest switching the model and table around: ``` SELECT m.saledate, m.ma, t.saledate, t.ma FROM example_db.demo_data.house_sales as t JOIN mindsdb.house_sales_model as m WHERE t.saledate > LATEST...
@alejandrovillegas did it work?
Ah, glad to hear! No, nothing urgent. I just saw this in my backlog and wondered what the status was :grin:, cheers
Another thing to consider @StpMax: We need to decide whether the `db.Predictor` class should be Lightwood-specific or not. I'm inclined to make it specific because things like the JsonAI are...
As for Lightwood: it's possible to install it for M1 machines using `miniconda + conda-forge`. This, in turn, enables MindsDB usage from said architecture, I was able to run both...
Status: `conda-forge` recipe for Lightwood is awaiting review. Once merged and published, we can modify MindsDB's setup procedure to install it from `conda-forge` rather than `pip` if we detect Mac...
@wangqinghuan we're currently working on this feature. There's already some progress in the AutoML engine, but we're still a bit away from implementing it in SQL. Will let you know...