Benchmark

One of the first implementations of the VFI algorithm is the one that is included in the Waikato Environment for Knowledge Analysis, in short WEKA, which is written in Java.

In the following table we can see that the proposed implementation achives better performance in most of the cases. In the Difference column if the Python implementation performs significant better the value is W, if the Java implementation performs significant better the value is L, otherwise we have a draw (D).

Dataset	Python	Java	Difference
abalone	58.08	43.06	W
acute-inflammation	99.83	98.17	D
acute-nephritis	95.83	91.67	D
arrhythmia	54.29	47.44	W
balance-scale	84.95	69.07	W
balloons	61.50	61.50	D
bank	71.17	56.23	W
blood	62.50	37.55	W
breast-cancer-wisc-diag	93.87	91.37	D
breast-cancer-wisc-prog	61.49	57.05	D
breast-cancer-wisc	96.42	92.86	W
breast-cancer	68.59	69.08	D
breast-tissue	62.06	64.52	D
car	81.38	70.09	W
cardiotocography-10clases	53.67	52.06	D
cardiotocography-3clases	77.30	75.55	D
chess-krvk	17.96	17.58	D
chess-krvkp	86.00	80.90	W
column_2C_weka	69.81	72.29	D
column_3C_weka	67.97	75.87	D
congressional-voting	56.71	52.45	W
conn-bench-sonar-mines-ro	74.27	57.51	W
contrac	48.54	42.40	W
credit-approval	85.33	78.84	W
cylinder-bands	70.98	66.39	D
dermatology	96.94	91.99	W
echocardiogram	79.82	81.46	D
ecoli	74.11	74.54	D
energy-y1	87.59	81.25	W
energy-y2	88.55	81.35	W
fertility	67.50	56.10	D
flags	57.98	48.14	W
glass	56.22	56.53	D
haberman-survival	62.38	57.83	D
heart-cleveland	56.04	43.09	W
heart-hungarian	85.39	83.37	D
heart-switzerland	22.21	25.09	D
heart-va	31.70	33.95	D
hepatitis	80.52	69.64	D
ilpd-indian-liver	53.53	46.67	W
ionosphere	80.92	94.50	L
iris	95.73	96.07	D
led-display	70.82	73.53	L
lenses	86.50	75.17	D
letter	64.91	61.17	W
libras	59.89	58.58	D
low-res-spect	75.44	76.54	D
lung-cancer	53.92	54.25	D
lymphography	80.52	80.32	D
magic	74.63	66.50	W
mammographic	79.05	61.82	W
molec-biol-promoter	88.82	67.65	W
molec-biol-splice	89.86	39.76	W
mushroom	90.17	78.85	W
musk-1	75.92	79.73	D
musk-2	74.56	79.06	L
nursery	73.06	73.67	D
oocytes_merluccius_nucleu	61.22	44.74	W
oocytes_merluccius_states	85.58	84.64	D
oocytes_trisopterus_nucle	58.57	55.75	D
oocytes_trisopterus_state	74.65	70.34	W
ozone	71.03	56.47	W
page-blocks	68.18	87.24	L
parkinsons	75.78	70.36	D
pima	72.44	62.57	W
pittsburg-bridges-MATERIA	78.32	78.09	D
pittsburg-bridges-REL-L	59.58	53.74	D
pittsburg-bridges-SPAN	58.09	66.69	D
pittsburg-bridges-T-OR-D	72.65	62.11	D
pittsburg-bridges-TYPE	51.75	41.69	D
planning	49.03	41.98	D
plant-margin	45.56	62.45	L
plant-shape	37.99	37.69	D
plant-texture	29.26	55.48	L
post-operative	42.56	32.33	D
primary-tumor	38.97	37.55	D
ringnorm	91.76	92.06	D
seeds	89.43	88.19	D
semeion	79.52	78.81	D
spambase	63.84	68.97	D
statlog-australian-credit	59.57	55.32	D
statlog-german-credit	71.04	63.36	W
statlog-heart	83.78	79.89	D
statlog-image	80.41	77.60	W
statlog-vehicle	61.32	53.18	W
steel-plates	55.22	55.09	D
synthetic-control	93.08	93.03	D
teaching	46.59	42.39	D
tic-tac-toe	66.71	70.46	D
titanic	75.52	77.60	L
trains	80.00	90.00	D
twonorm	95.51	55.51	W
vertebral-column-2clases	69.81	72.29	D
vertebral-column-3clases	68.00	75.65	D
wall-following	67.07	92.21	L
waveform-noise	76.46	56.25	W
waveform	76.61	54.56	W
wine-quality-red	37.54	35.13	D
wine-quality-white	27.43	27.84	D
wine	94.71	96.16	D
yeast	44.44	50.52	L
zoo	94.95	90.18	D
(L/D/W)			(9/59/34)