Table 1 Results of the FSL models on...

Table 1

Results of the FSL models on test sets

Input sequence length	Accuracy	95% C.I.	ROC-AUC
2 (2012–2013)	0.802	[0.778, 0.794]	0.880
3 (2012–2014)	0.826	[0.792, 0.816]	0.897
4 (2012–2015)	0.836	[0.799, 0.821]	0.910
5 (2012–2016)	0.862	[0.831, 0.846]	0.926
6 (2012–2017)	0.859	[0.844, 0.853]	0.924
7 (2012–2018)	0.851	[0.823, 0.838]	0.924
8 (2012–2019)	0.850	[0.832, 0.845]	0.924
9 (2012–2020)	0.844	[0.826, 0.841]	0.914

Input sequence length	Accuracy	95% C.I.	ROC-AUC
2 (2012–2013)	0.802	[0.778, 0.794]	0.880
3 (2012–2014)	0.826	[0.792, 0.816]	0.897
4 (2012–2015)	0.836	[0.799, 0.821]	0.910
5 (2012–2016)	0.862	[0.831, 0.846]	0.926
6 (2012–2017)	0.859	[0.844, 0.853]	0.924
7 (2012–2018)	0.851	[0.823, 0.838]	0.924
8 (2012–2019)	0.850	[0.832, 0.845]	0.924
9 (2012–2020)	0.844	[0.826, 0.841]	0.914

Note(s): Each model has been tested by using 1500 samples of the same sequence length. The table reports the accuracy of the best model (over cross-validation) on the bankruptcy binary classification, followed by the confidence interval of accuracy over the results of cross validation and the ROC-AUC score of the best performing model

Source(s): Table created by the authors

[ViewLarge]

Sharing Unavailable