Flaml meets DoubleML - Comparing AutoML tuning

[sauravbania]

In this notebook, we compare AutoML tuning methods using the FLAML library. We generate synthetic data with the make_plr_CCDDHNR2018 function. The analysis includes:

1.	AutoML Manual Tuning: Hyperparameters are manually set for tuning.
2.	AutoML API Tuning: The FLAML API automatically optimizes hyperparameters.
3.	Dummy Models: Simple mean-based models as a baseline.

We assess model performance using the DoubleML framework, focusing on metrics like Mean Squared Error (MSE) and coefficients with confidence intervals. Results are saved and visualized to compare the effectiveness of each tuning approach.