README.md

Federated Experiments

Communication Rounds vs Grid Size (Figure 17)

python comm.py --dataset blood  --feature Monetary
python comm.py --dataset cancer --feature area1
python comm.py --dataset ecoli  --feature lip
python comm.py --dataset ecoli  --feature chg
python comm.py --dataset house  --feature Street
python comm.py --dataset house  --feature YrSold

Plots are saved in ../img/federated/{power}-{dataset}-{feature}.pdf.

Parameter	Description	Values
power	Power transform method	boxcox or yeojohnson
dataset	Dataset name	adult, bank, credit, blood, cancer, ecoli, house
feature	Feature index or name	Integer (e.g., 0, 1, ...) or string (e.g., feature name)

---

Simulate Federated Power Transform

python simulate.py --power yeojohnson --dataset blood --full_output 1

Results are printed in the terminal.

Parameter	Description	Values
power	Power transform method	boxcox or yeojohnson
dataset	Dataset name	adult, bank, credit, blood, cancer, ecoli, house
feature	Feature index or name	Integer (e.g., 0, 1, ...) or string (e.g., feature name)
var_comp	Variance computation method	pairwise or naive
optimize	Optimization method	brent or grid
n_points	Number of points for grid search	Integer (e.g., 20, 50)
n_clients	Number of clients	Integer (e.g., 10, 100)
full_output	Whether to return full output	0 or 1
n_reps	Number of repetitions	Integer (e.g., 1, 3)
print_output	Print output or not	0 or 1