Method Verification

[yoogawu]

Hello, I would like to ask how the accuracy of each method is verified? Is there a comparison with the original text?

[yoogawu]

I'm currently reading and trying to reproduce the sasfunclust method, so I wanna check this issue.

[Barry57]

Below is our comparison of each method with the simulation cases reported in the papers. We believe our code can reproduce the published results; any small deviations come from the randomness inherent in simulating new data and from minor differences between Python and R implementations. You can call the SAS algorithm and experiment with different settings yourself—every parameter in our implementation follows the specifications given in the original articles.

Category	Method	Case	Metric	Original	BiFuncLib
Functional biclustering	FunFEM	The Kneading dataset	Clustering accuracy	70.97%	71.25%
Functional biclustering	FunLBM	Scenario B	ARI (R+C)	R:≈0.95 / C:≈0.97	R:0.983(0.06) / C:0.971(0.08)
Functional biclustering	FunCC	Case B	Correctly identify	3/3	3/3
Functional biclustering	FunPF	Example 1, N=30	ARI (R+C)	R:0.911(0.28) / C:0.910(0.28)	R:0.909(0.28) / C:0.911(0.29)
Local clustering	FunSparse	Simulation 2B, d=0.7	CER	0.173(0.12)	0.174(0.11)
Local clustering	FunSAS	Scenario 1, σₑ=2	ARI	≈0.90–0.95	0.926(0.15)
Local clustering	FunLocal	Example 1, σ=1, n=100	ARI	0.978(0.02)	0.976(0.02)
Scalar biclustering	Bimax	Scenario 1, σ=1	Module Recovery	≈0.85–0.9	0.894(0.17)
Scalar biclustering	SSVD	Case 1	CER (R+C)	R:1.01% / C:0.24%	R:1.16% / C:0.19%
Scalar biclustering	CVX	N_b=16, σ=1.5	ARI	0.952	0.947(0.00)

[yoogawu]

Thx for your reply, I'm trying.