lightwood
lightwood copied to clipboard
mindsdb
Lightwood is Legos for Machine Learning.
The objective is to be able to add them in `~/lightwood_modules` for better user experience.
Title says it all
Additional insights regarding the total sum of the predicted median, min and max in numerical forecasts.
Implement a prediction argument that enables higher frequency forecasts. For example, if we called a predictor trained on a daily dataset with `HORIZON=3` like this: ```python predictions = predictor.predict(args={'freq': 'H'})...
Once #989 is merged. With this change in place, it doesn't make sense anymore to define offsets when predicting. Instead, we should ask for `infer_mode` to be set through a...
We should add a new method to the predictor interface for arbitrary metric computation given a dataframe with values for the target column. ```python pdef = ProblemDefinition.from_dict({...}) json_ai = json_ai_from_problem(train_df,...
From MindsDB. If a user inputs this query: ```sql SELECT m.saledate as date, m.ma as forecast, m.ma_explain FROM mindsdb.house_sales_model as m JOIN example_db.demo_data.house_sales as t WHERE t.saledate > LATEST AND...
Currently it seems that on a monthly dataset, the interval is estimated as 30/31 days instead of monthly. Based on the input data, is it possible to identify the interval...
We should add docstrings where missing for all analysis blocks, for reference in the autogenerated docs.
So that `global_insights` will include this bit: 