lightwood
lightwood copied to clipboard
mindsdb
Lightwood is Legos for Machine Learning.
Two main approaches: - Disable optuna hyperparam search past some threshold. - Opt for a recursive strategy instead of direct (i.e. same regressor trained for all timesteps v/s one for...
Following recent changes to the analysis block parser, we should be able to specify things like encoders and mixers (ideally all possible modules, though) __without__ the need for initial args....
[Reference issue](https://github.com/mindsdb/mindsdb/issues/2600)
Often additional features proposed to users include day_of_week, weekend, hour_of_day etc. In the encoding of the time variable, could we add these columns as additional inputs?
[Reference issue](https://github.com/mindsdb/mindsdb/issues/2600) Trigger the behavior in #958 automatically if partitions are not valid (i.e. duplicate time entries but different target readings). Important to clearly communicate default policies here because it...
[Reference issue](https://github.com/mindsdb/mindsdb/issues/2600) Internally sample and derive new time granularity given `tss.resample` value, and aggregates all other columns as per `tss.reduce`. The latter may be difficult to do for multivariate framings,...
This will be useful for reference because internally columns can be dropped or added.
Important logic, we should unittest it with real-world examples and generated data
