search

mljar / mljar-supervised

Python package for AutoML on Tabular Data with Feature Engineering, Hyper-Parameters Tuning, Explanations and Automatic Documentation
https://mljar.com
MIT License
2.99k stars 400 forks source link

Set parameter/constraint for algorithm, for example: change constraint for Linear #452

Open jmrichardson opened 3 years ago

jmrichardson commented 3 years ago

Hi, In this section of code there appears to be a constraint on the number of rows and columns used for the Linear model:

https://github.com/mljar/mljar-supervised/blob/013094ece2d4f76411fd5a014946ab1c7d31842c/supervised/base_automl.py#L738

Is there a way to make this user defined? I am using Rocket which by default creates 20K features (convolutional kernels). Using a linear model works well for this and wanted to see how mljar performs. I could hack the code but would be great to have a setting to remove the constraint if possible.

pplonski commented 3 years ago

@jmrichardson for now the best way is to hack the code. I will add this issue as a new feature to set any parameter/constraint for the algorithms.