Application of Feedforward Neural Network and SPT Results in the Estimation of Seismic Soil Liquefaction Triggering

Table 10

Comparison with other empirical formulas and ML models.

Author	Model	Number of samples	Dataset	Wrong predicted	Accuracy (%)
Seed and Idriss (1971) [12]	Empirical formula	296	Entire dataset	24/296	91.89
Liao and Whitman (1981) [40]	Empirical formula	296	Entire dataset	28/296	90.54
Youd et al. (2003) [14]	Empirical formula	296	Entire dataset	23/296	92.23
Robertson and Fear (1995) [13]	Empirical formula	486	Training set	83/388	78.61
			Testing set	21/98	94.59
			Entire dataset	104/486	78.60
Boulanger and Idriss (2016) [34]	Empirical formula	486	Training set	48/388	87.63
			Testing set	9/98	90.82
			Entire dataset	57/486	88.27
Shahri and Moud (2020) [22]	ICA-MOGFFN	296	Entire dataset	20/296	93.24
	MOGFNN	296	Entire dataset	25/296	91.55
Sarat Kumar Das (2011) [21]	MGGP	227	Entire dataset	30/227	86.78
This study	MVR	486	Training set	206/388	46.91
			Testing set	15/98	84.69
			Entire dataset	221/288	54.53
	RS-L2-FNN		Training set	36/388	90.72
			Testing set	11/98	88.78
			Entire dataset	47/486	90.33