Predicting citation count of Bioinformatics papers within four years of publication

Departamento de Inteligencia Artificial, Universidad Politécnica de Madrid. 28660 Madrid, Spain

Predicting citation count of Bioinformatics papers

within four years of publication

RESULTS

We will build two different types of prediction models: Global models and Specific models. Global models attempt to predict the number of citations received by an article within each of one the next four years after publication, using information of all papers published in Bioinformatics journal during three years, between January 1, 2005 and December 31, 2007. Specific models have the same objective as global models, but in this case, they use the information related to articles published within a specific Bioinformatics journal section.

Paradigms used in this work are: Bayesian networks (naiveBayes and K2), logistic regression, decision trees (C4.5) and k-nearest neighbors (5-NN).

General models

Percentage of well classified cases within the next four years after publication and confusion matrices. (Table)

Specific models

Average accuracy within the four prediction years by each section and paradigm. (Table, Figure).

Average accuracy within the nine publication journal sections by each prediction year and paradigm. (Table, Figure).

Structure models:

Bayesian networks (naiveBayes): Section 1 - First-year Section 2 - First-year

Bayesian networks (K2): Section 2 - Fourth-year Section 4 - First-year

Decision trees (C4.5) Section 4 - Second-year Section 8 - Fourth-year

Accuracy and standard deviation. Numbers in boldface represent an average success rate better than 95%.

		Section 1	Section 2	Section 3	Section 4	Section 5	Section 6	Section 7	Section 8	Section 9
First year	NB	96.6 5.42	91.9 12.1	94.0 4.42	95.6 8.45	93.2 4.73	100 0.00	96.3 3.58	97.3 3.42	96.9 3.32
	K2	95.5 5.74	91.9 12.1	94.0 3.26	97.8 6.32	92.2 4.09	96.4 10.5	96.3 3.55	95.3 4.50	96.3 3.20
	LR	98.9 3.53	86.5 17.7	90.8 2.55	95.7 8.46	94.2 5.06	92.9 14.0	94.7 4.36	92.0 4.26	91.9 4.81
	C4.5	92.0 5.42	86.5 13.2	90.1 2.22	89.1 10.5	90.3 0.48	82.1 17.6	89.5 2.53	86.7 3.19	88.2 3.59
	K-NN	92.0 5.42	86.5 13.2	89.4 0.22	89.1 10.5	90.3 0.48	82.1 17.6	89.5 0.00	86.7 0.00	87.0 1.65
------------	---------	--------------------	---------------------	----------------------	----------------------	----------------------	----------------------	----------------------	----------------------	----------------------
Second year	NB	82.9 12.3	89.2 21.9	86.2 6.23	84.8 14.1	86.4 6.59	89.3 16.1	85.8 6.65	90.7 7.20	87.0 8.52
	K2	69.3 15.4	73.0 23.3	75.3 7.39	78.3 15.3	70.9 14.5	67.9 24.6	80.0 7.94	68.0 12.1	80.7 11.5
	LR	96.6 5.43	94.6 12.4	88.7 5.42	95.7 9.62	95.1 6.93	92.9 14.0	95.3 5.87	94.7 7.60	93.2 6.17
	C4.5	56.8 11.4	37.8 11.3	53.7 6.06	58.7 22.8	61.2 14.9	57.1 30.6	58.9 8.03	54.0 6.90	56.5 9.60
	K-NN	53.4 9.34	73.0 20.1	62.2 8.94	50.0 17.1	63.1 15.5	60.7 24.6	57.4 7.91	72.7 15.5	60.2 9.35
------------	---------	--------------------	---------------------	----------------------	----------------------	----------------------	----------------------	----------------------	----------------------	----------------------
Third year	NB	90.0 14.1	80.0 21.9	85.4 6.81	88.9 22.5	95.1 6.34	85.0 24.1	86.5 7.14	89.3 11.0	84.0 11.7
	K2	72.0 21.5	72.0 25.1	70.8 7.23	63.0 28.1	73.2 14.4	90.0 21.1	71.6 14.2	76.7 15.4	77.0 14.9
	LR	94.0 9.78	100 0.00	89.6 7.74	92.6 14.0	93.9 6.66	100 0.00	95.7 5.09	93.2 6.70	93.0 8.20
	C4.5	50.0 14.1	48.0 12.3	66.7 7.86	48.1 21.4	54.9 9.78	70.0 35.0	58.2 9.23	58.3 11.1	57.0 11.6
	K-NN	64.0 15.8	52.0 12.3	54.7 8.70	92.6 21.1	58.5 14.2	75.0 26.3	58.9 9.12	65.0 11.4	60.0 6.70
------------	---------	--------------------	---------------------	----------------------	----------------------	----------------------	----------------------	----------------------	----------------------	----------------------
Fourth year	NB	91.7 21.1	80.0 24.1	86.0 4.53	90.5 18.0	90.6 9.93	81.8 42.2	93.9 10.6	87.0 12.0	90.6 13.9
	K2	75.0 33.5	66.7 34.9	71.0 16.9	85.7 21.9	77.4 22.7	81.8 42.1	75.6 15.7	83.3 9.90	88.7 13.8
	LR	91.7 18.0	66.7 45.9	93.5 10.8	85.7 24.9	88.7 15.7	63.3 51.6	95.1 12.1	81.5 19.8	90.6 19.4
	C4.5	33.3 24.1	26.7 35.4	55.1 15.8	66.7 27.7	41.5 14.1	63.6 47.4	59.8 10.2	59.3 18.6	69.8 11.1
	K-NN	83.3 24.1	53.3 40.8	72.9 14.8	66.7 27.7	58.5 16.3	45.4 49.7	63.4 19.9	68.5 9.30	71.7 8.80

HOME DATA RESULTS EXPLOITING BEST MODELS