Prediction of Flight Time Deviation for Lithuanian Airports Using Supervised Machine Learning Model

Table 3

The arrival flight estimation results (expressed as a percentage).

Algorithm	Dataset	Measures (%)
		Sensitivity/recall	Precision	Specificity	F measure	Accuracy
Probabilistic neural network	Testing	78.87	76.63	90.28	77.23	76.63
	New data	36.56	33.21	74.17	34.55	33.54
Multilayer perceptron	Testing	56.73	47.43	73.01	50.77	47.43
	New data	39.45	32.15	66.87	35.17	31.78
Decision trees	Testing	66.56	63.11	85.78	64.42	63.11
	New data	61.12	57.84	83.55	59.12	57.83
Random forest	Testing	81.27	77.40	89.82	78.40	77.40
	New data	64.20	58.82	82.77	60.41	58.97
Tree ensemble	Testing	81.00	78.33	90.93	79.04	78.33
	New data	69.72	66.15	86.47	67.14	66.19
Gradient boosted trees	Testing	89.45	88.59	96.00	88.85	88.59
	New data	81.98	80.75	93.28	81.09	80.90
Support vector machine	Testing	91.98	32.70	6.47	46.41	32.70
	New data	96.42	32.01	4.26	47.50	32.73