Graphics Reference
In-Depth Information
•
The imputation methods which fill in the MVs outperform the case deletion (IM
method) and the lack of imputation (DNI method).
•
There is no universal imputation method which performs best for all classifiers.
In Sect.
4.6.1
we have analyzed the influence of the imputation methods in the data
in respect to two measures. These two measures are the
Wilson's noise ratio
and the
average MI difference
. The first one quantifies the noise induced by the imputation
method in the instances which contain MVs. The second one examines the increment
or decrement in the relationship of the isolated input attributes with respect to the
class label. We have observed that the CMC and EC methods are the ones which
introduce less noise and maintain the MI better.
According to the results in Sect.
4.6.2
, the particular analysis of the MVs treat-
ment methods conditioned to the classification methods' groups seems necessary.
Thus, we can stress the recommended imputation algorithms to be used based on
the classification method's type, as in the case of the
FKMI
imputation method for
the Rule Induction Learning group, the
EC
method for the black-boxes modelling
Models and the
MC
method for the Lazy Learning models. We can confirm the pos-
itive effect of the imputation methods and the classifiers' behavior, and the presence
of more suitable imputation methods for some particular classifier categories than
others.
References
1. Acuna, E., Rodriguez, C.: Classification, Clustering and Data Mining Applications. Springer,
Berlin (2004)
2. Atkeson, C.G., Moore, A.W., Schaal, S.: Locally weighted learning. Artif. Intell. Rev.
11
,
11-73 (1997)
3. Aydilek, I.B., Arslan, A.: A hybrid method for imputation of missing values using optimized
fuzzy c-means with support vector regression and a genetic algorithm. Inf. Sci.
233
, 25-35
(2013)
4. Azim, S., Aggarwal, S.: Hybrid model for data imputation: using fuzzy c-means and multi
layer perceptron. In: Advance Computing Conference (IACC), 2014 IEEE International,
pp. 1281-1285 (2014)
5. Barnard, J., Meng, X.: Applications of multiple imputation in medical studies: from aids to
nhanes. Stat. Methods Med. Res.
8
(1), 17-36 (1999)
6. Batista, G., Monard, M.: An analysis of four missing data treatment methods for supervised
learning. Appl. Artif. Intell.
17
(5), 519-533 (2003)
7. Bezdek, J., Kuncheva, L.: Nearest prototype classifier designs: an experimental study. Int. J.
Intell. Syst.
16
(12), 1445-1473 (2001)
8. Broomhead, D., Lowe, D.: Multivariable functional interpolation and adaptive networks.
Complex Systems
11
, 321-355 (1988)
9. van Buuren, S., Groothuis-Oudshoorn, K.: MICE: multivariate imputation by chained equa-
tions in r. J. Stat. Softw.
45
(3), 1-67 (2011)
10. le Cessie, S., van Houwelingen, J.: Ridge estimators in logistic regression. Appl. Stat.
41
(1),
191-201 (1992)
11. Chai, L., Mohamad, M., Deris, S., Chong, C., Choon, Y., Ibrahim, Z., Omatu, S.: Inferring
gene regulatory networks from gene expression data by a dynamic bayesian network-based
model. In: Omatu, S., De Paz Santana, J.F., González, S.R., Molina, J.M., Bernardos, A.M.,
