Financial Innovation

Table 9 Feature importance for logistic regression models with undersampling

From: Detecting DeFi securities violations from token smart contract code

LR1		LR2		LR3		LR4		LR5		LR6		LR7
Feature	Imp.	Feature	Imp.	Feature	Imp.	Feature	Imp.	Feature	Imp.	Feature	Imp.	Feature	Imp.
SWAP15	0.396	PUSH30	0.637	CALLVALUE	0.043	SWAP7	0.515	BALANCE	0.260	PUSH30	0.701	PUSH30	0.671
RETURN	0.340	SWAP15	0.514	CODESIZE	0.484	CALLVALUE	0.514	CALLER	0.023	SWAP15	0.389	LOG4	0.215
PUSH30	0.299	CODESIZE	0.443	EXP	0.020	PUSH9	0.430	CALLVALUE	0.058	BALANCE	0.215	PUSH21	0.168
CALLVALUE	0.287	PUSH31	0.146	LOG1	0.070	RETURN	0.427	LOG1	0.145	LOG1	0.165	RETURN	0.140
LOG1	0.253	LOG1	0.102	PUSH30	0.592	SWAP15	0.382	PUSH30	0.706	PUSH31	0.088	SLT	0.118
PUSH9	0.248	PUSH9	0.091	PUSH31	0.184	PUSH31	0.377	PUSH31	0.124	SWAP7	0.064	PUSH9	0.083
PUSH31	0.233	CALLVALUE	0.050	PUSH9	0.025	PUSH30	0.328	PUSH9	0.038	CALLVALUE	0.061	DUP2	0.022
EXP	0.230	EXP	0.022	SWAP15	0.590	LOG1	0.271	RETURN	-0.022	CALLER	0.012	MLOAD	0.007
SWAP7	0.197	RETURN	-0.030			CALLER	0.266	SWAP15	0.450	PUSH9	0.003	SHL	-0.031

Back to article page