Table 5.

Performance of the cell quality classifier. The performance of the cell quality classifier was close to that of the experts. In addition, the performance for Edom was lower than that for the other cell lines because of the cell line dependency.

	Class	Three cell lines
		Number	Accuracy	Precision	Recall	$F$-measure
Training (three cell lines)	Poor	266	0.810	0.768	0.820	0.793
	Mod	172	0.810	0.693	0.605	0.646
	Good	162	0.870	0.753	0.772	0.762
Test (three cell lines)	Poor	135	0.840	0.832	0.807	0.820
	Mod	87	0.823	0.707	0.667	0.686
	Good	78	0.897	0.770	0.859	0.812
Test (MRC5)	Poor	36	0.860	0.775	0.861	0.816
	Mod	39	0.890	0.938	0.769	0.845
	Good	25	0.890	0.750	0.840	0.792
Test (Edom)	Poor	53	0.800	0.902	0.698	0.787
	Mod	28	0.750	0.548	0.607	0.576
	Good	19	0.870	0.607	0.895	0.723
Test (201B7)	Poor	46	0.860	0.820	0.891	0.854
	Mod	20	0.830	0.579	0.550	0.564
	Good	34	0.930	0.935	0.853	0.892