Received November 30, 2017, accepted December 31, 2017, date of publication January 4, 2018,

date of current version February 28, 2018.

Digital Object Identifier 10.1109/ACCESS.2018.2789428

Class Weights Random Forest Algorithm for

Processing Class Imbalanced Medical Data
MIN ZHU 1,2 , JING XIA1 , XIAOQING JIN3 , MOLEI YAN3 , GUOLONG CAI3 ,
JING YAN3 , AND GANGMIN NING1
1 Department of Biomedical Engineering, Zhejiang University, Hangzhou, Zhejiang 310027, China
2 Guizhou Key Laboratory of Agricultural Bioengineering, Guizhou University, Guiyang 550025, China
3 Zhejiang Hospital, Hangzhou 310058, China

Corresponding authors: Jing Yan (zjicu@vip.163.com) and Gangmin Ning (gmning@zju.edu.cn)

This work was supported in part by the National Natural Science Foundation of China under Grant 81271662, in part by the Ministry of the
Science and Technology of China under Grant 2014DFT3050, and in part by the Department of Science and Technology of Zhejiang
Province under Grant 2013C03049-2.

ABSTRACT The classification in class imbalanced data has drawn significant interest in medical application.
Most existing methods are prone to categorize the samples into the majority class, resulting in bias,
in particular the insufficient identification of minority class. A kind of novel approach, class weights random
forest is introduced to address the problem, by assigning individual weights for each class instead of a single
weight. The validation test on UCI data sets demonstrates that for imbalanced medical data, the proposed
method enhanced the overall performance of the classifier while producing high accuracy in identifying both
majority and minority class.

INDEX TERMS Class imbalanced, random forest, weighted voting, class weights voting.

I. INTRODUCTION with larger distribution is named as the majority while the

Classification in medical diagnostics is able to aid in disease
diagnosis and predicts outcomes in response to the treatment.
Many efforts have been made to improve the classification
performance. For instance, in traditional methodology for
classification, logistic regression–based trichotomous classi-
fication tree was applied in diagnosing breast cancers [1].
Non-parametric empirical bayes algorithm was developed for
integrative genetic risk prediction of complex diseases with
binary phenotypes [2]. Hierarchical support vector machine-
based algorithm was employed in the EEG-based motor
imagery classification task [3]. Bionic algorithms were also
introduced in the classification of medical data. Self-adaptive
niche genetic algorithm with random forest was proposed
to build model for sepsis patient’s stratification [4], Clas-
sification rules were extracted by ant-miner algorithm and
thereby applied in diagnosing heart disease etc. [5]. Utilizing
neural network (CNN) algorithm, multimodal disease risk
prediction model was developed to predict the risk of chronic
disease in communities [6], and classify lung nodule by
unsupervised image features learning [7]. However, in the
practice of medical classification, data are usually class-
imbalanced [8], which means the distributions of classes are
not uniform [9]–[11]. In binary-classification cases, the class
other is named as the minority [12], [13]. Dealing with the
class-imbalanced data, conventional algorithms are prone to
consider tend to minority observation as noise or outliers and
ignore them in the classifying [12], thereby tend to classify
samples into the majority class [9], [14], [15]. Consequently,
the predictive accuracy for the minority class will be much
lower than that for the majority class [8], [16]–[20].
To solve the aforementioned problems, data usually need to
be processed to construct a balanced dataset [21]–[32]. How-
ever, in classification in medical diagnostics.., it is often desir-
able to retain as much data as possible. Therefore, the direct
application of data is widely employed. Classifier Combina-
tion is a practical method [18], [28]–[30], [33], [34], and the
ultimate goal of designing pattern recognition systems is to
achieve the best possible classification performance for the
task. This suggests that different classifier designs potentially
offering complementary information about the patterns to
be classified, which could be harnessed to improve perfor-
mance when dealing with class imbalanced data [35]–[40].
There were many popular algorithms concerning about
Classifier Combination; such as Bayesian [41], [42],
Dempster–Shafer [43]–[47], Fuzzy Integral [48], [49], and
Voting Methods [50]–[57]. The Voting Strategy which is

2169-3536
2018 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 6, 2018 Personal use is also permitted, but republication/redistribution requires IEEE permission. 4641
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

built by Bayesian and Novel Naive Bayes is proposed [58]. of each classifier into an ultimate prediction [87]. Different
A Combination Method based on the Dempster–Shafer weights per class are obtained from the empirical error of
Theory of Evidence is built [59], an Evidence-based Com- different classifiers. The algorithm assigns individual weights
bining Classifiers method is proposed for brain signal anal- for each class instead of a single weight and focuses on
ysis [60]. Dempster’s Rule is combined multiple classifiers the problem of effective identification for the minority class.
using for text categorization [61]. Dempster-Shafer Fusion It can improve the recognition performance for minority class
is combined models of prostate cancer [62]. Fuzzy Integral while maintaining those for majority class.
and Genetic Algorithms are combined multiple neural net-
II. METHOD
work classifiers [63]. Bayesian and Dempster–Shafer meth-
ods require a priori knowledge, and the calculation of the The algorithm, CWsRF is proposed; there are three proce-
fuzzy measure function is very complex [58]–[63]. To find dures as fellows: Building RF Model, Building CWsV and
a more convenient way to build the model, we chose Voting Classifying Votes; as shown in Fig. 1.
Method, which uses a unique method to integrate classifiers.
Furthermore, the calculation is simple and does not require
an auxiliary combiner or preset function, it is not limited to
decision trees with axis-parallel splits, and it is applicable to
any type of classifiers [64].
The most popular method of voting is major voting, which
based on random forest (RF) algorithm. Random forest clas-
sifiers can achieve high accuracy in data classification com-
pared with many standard classification methods [65]–[68],
and it can minimize the overall classification error rate and
has the ability to handle class imbalanced data [4], [56], [69].
However, when the imbalanced rate increases (e.g., 15%),
the classification ability is weakened [56], [70], [71], it is
because that each classifier has the same weight when the
classifiers are combined [70], [72]. Therefore, to solve this
problem, a method of combining classifiers with different
weights is proposed. Weight voting random forest (WRF),
an algorithm of getting the decisions of each classifier is
multiplied by a weight to reflect the individual confidence
of these decisions [73]–[75]. A weighted majority voting
method which is based on class-conditional independence of
FIGURE 1. The framework: CWsRF.
the classifier outputs is proposed [76]. Endogenous voting
weights is mentioned for elected representatives and redis- A. THE FRAMEWORK OF CWsRF
tricting [64]. Power in Weighted Voting Games with Super- RF takes the same weight for different classes of samples
Increasing Weights is analyzed [77]. A random forests (RF) and ensembles by major voting, which makes the classifiers
method with weighted voting for the task of anomaly detec- sensitive to the majority class(MAJ), and the classification
tion is presented [72], [78]. The internal out-of-bag (OOB) performance of the minority class(MIN) is decreased when it
error metric is used as a tree-weight in RF [79]. However, faced with imbalanced data. Therefore, an approach of CWsV
these approaches have not dramatically improved predictive is designed to distinguish MAJ and MIN.
ability. Each classifier only has a single weight when the There are three procedures: 1) Building RF Model: the
classifiers are combined [80], [81], which is not adequately vote of RF is obtained. 2) Building CWsV Method: it is the
distinguish between the majority class and the minority class; key procedure of the proposed algorithm (CWsRF) is mainly
the class weights (CWs), which refer to the different weights being introduced. It has two steps: a) the most important step
per class of each classifier are not obtained. Based on this is different weights per class were calculated; each classifier
observation, classifiers require multiple weights, and increas- has two different weights (minority and majority weight);
ing the CWs of each classifier has the potential to improve b) the votes of the samples are calculated. 3) Classifying
the overall predictive performances [82]. Therefore, CWs Votes: the improved votes are classified using the threshold:
should be assigned to better represent the classifier’s ability Aggregating Probability (AP).
to distinguish the difference between the majority class and
the minority class [21], [83]–[86]. B. THE PROCEDURES OF CWsRF
Therefore, we proposed a class weights voting (CWsRF) 1) BUILDING RF MODEL
algorithm based on random forest algorithm (RF), which Traditional RF is building to obtain the vote of classifiers.
contains an approach (CWsV) and trains a collection of clas- vi,j,c , v_tei,j,c , the labels denoting the jth classifier in the
sifiers in different weights per class to combine the output ith sample of the training set and test set need to be first

4642 VOLUME 6, 2018

M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

TABLE 1. vi ,j ,c : The labels that classified by jth classifier in ith sample Step 2 (Calculating ACCj,c ): ACCj,c , the accuracy of
(train set).
the cth class in the jth classifier, is the difference accuracy
between each class per classifier. Thus, each classifier has two
ACCj,c values. nMAJ is the number of MAJ samples; nMIN is
the number of MIN samples. HMIN are the classifiers when
vi,j,c ∈ MIN . HMAJ are the classifiers when vi,j,c ∈ MAJ .
PnMAJ H
i=1 PHMAJ
scorei,j,MAJ
j=1
ACCj,MIN = (2)
TABLE 2. v _tei ,j : The labels that classified by jth classifier in ith sample nMAJ
(test set). PnMIN H
i=1 PHMIN score
j=1 i,j,MIN
ACCj,MAJ = (3)
nMIN
Step 3 (Calculating wj,c ): wj,c , the weights of each classi-
fier per class, are calculated to obtain new voting results based
on ACCj,c . They also have two values.
wj,MIN = ACCj,MIN (4)
obtained, they have one statue, either MIN or MAJ, as shown wj,MAJ = ACCj,MAJ (5)
in TABLE 1 and TABLE 2. j is the number of classifiers. i is
(2) Calculating Votes
the number of samples, c belongs to the MIN or MAJ, and xi
Train set and test set’s votes are obtained, each sample
is the original label of samples of the training set.
gets two votes, there are two steps: calculating vtri,j vtei,j ;
calculating vtraini,c , vtest i,c
2) BUILDING CWsV
vtri,j vtei,j are concert to train set, test set, they are the votes
CWsV, Key procedures of CWsRF will be presented in two in the jth classifier to the ith sample as wj,c are promoted. They
steps, as shown in Fig. 2. have one statue, either MIN or MAJ
vtri,j,c = vi,j,c × wj,c (6)
vtei,j,c = v_tei,j,c × wj,c (7)
vtraini,c , vtest i,c are concert to train set, test set; they are
the total votes to ith sample in MIN and MAJ.
XHc
vtraini,c = vtr i,j,c (8)
j=1
XHc
vtest i,c = vtei,j,c (9)
j=1

3) CLASSIFYING VOTES
Threshold voting is used to instead of major voting in tradi-
tional RF. There are two steps
(1) Calculating vtrnewi,j,c
vtr_newi,j,c is the vote of ith sample in different j threshold
weight, which has one statue, either MIN or MAJ.

FIGURE 2. Different weights per class of classifiers. Algorithm 1 Pseudo-Code of Aggregating Probability
for i ∈ [1 . . . n]
The classification capability of each classifier is often used for j ∈ [−H . . . H]
to evaluate the weight of a classifier; therefore the classifier’s if vtraini,MIN − vtraini,MAJ > j
prior accuracy (ACC) is used to measure the different weight Output: vtr_newi,j,c ← MIN (c ∈ MIN)
per class (W ∝ ACC); and the weights were used to calculate else
votes. Output: vtr_newi,j,c ← MAJ (c ∈ MAJ)
(1) Calculating Different Weight per Class end for
Step 1 (Calculating scorei,j,c ): scoreij,c , the score of each end for
classifier for each sample, they have one statue, either 1 or 0.
The equation is given as follows:

vtr_newi,j,c is shown in TABLE 3:
scorei,j,c = 1 vi,j,c == xi = c (1) (2) Obtaining AP

VOLUME 6, 2018 4643

 M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

TABLE 3. vtr_newi ,j ,c : The result of vote in ith sample between TABLE 4. Data basic information.
different J.

A best AP value is obtained through max AUC. AUC is

widely used in evaluating the performance of classification
of imbalanced data sets: C. FLOWCHART
PnMIN PnMIN According to the above analysis, the flowchart of CWsRF
i=1 (ri −i) (ri ) − nMIN ×(n2MAJ +1) is presented, shown in Fig. 3. The weights of the classifiers
AUC = = i=1 (10)
nMIN × nMAJ nMIN × nMAJ between MIN and MAJ are distinguished.
nMIN and nMAJ are the number of minority and majority cases,
III. DATASET DESCRIPTION AND EXPERIMENTAL SETUP
and ri is an ordinal number in the sorting table of the ith
A. DATASET DESCRIPTION
minority case.
To evaluate the different performances, five databases are
Then,

the AUCs of different j are calculated, argmax used, four from the UCI Machine Learning [SPECTF
AUCj is obtained, the current value j is recorded, and
(SP), Mammographic (MA), Diagnostic Wisconsin Breast
AP = j, the optimal AP is obtained. After that, AP and vtesti,c
Cancer (WD), Colic (CO)] and one collected by Zhejiang
are used to determine the result of classifying, as shown
Hospital [Osteoporosis (OST)]. Their size, type, and source
in Fig. 3.
are shown in TABLE 4.
The SP database contains cardiac single proton emission
computed tomography images collected during the myocar-
dial perfusion diagnostic process [88]. The MA database
contains information on the discrimination of benign and
malignant mammographic masses. The WDBC database is
the Diagnostic Wisconsin Breast Cancer Database. The Colic
database contains horse-colic data. OST contains osteo-
porosis data. The imbalanced information contained in the
included datasets is summarized in TABLE 5.

TABLE 5. Imbalanced information of data.

In medical data collection, the class imbalanced rate of

data varies because of different incidences. For example,
according to GLOBOCAN statistics, female breast cancer
incidence rates range from 25.6 cases per 100,000 females in
Thailand to 95.3 per 100,000 in the Netherlands [88], and in
the affected population, the mortality rate is approximately
10% to 25%. Osteoporotic fracture in China ranges from
40-50% in women and is 20% for men [89]. The inci-
dence varies greatly among different diseases, and in addi-
tion, in machine learning fields, the class imbalanced ratio
FIGURE 3. Flowchart. between the minority class and the majority class is not less

4644 VOLUME 6, 2018

M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data
TABLE 6. The results of AUC, F1_score, Recall for different IRs where IR = 25%, 20%, 15%, and 10%.
than 20-25% [90], [91]. Therefore, the content above was
combined to show the versatility of the algorithm, and the
datasets were altered by different imbalanced rates (IRs):
the minority class was set to 25%, 20%, 15%, or 10% of
the majority class, respectively. Moreover, the IR and the
incidences of the datasets used in this paper were matched.
These findings show that the algorithm presented here has
high practical significance.
B. PARAMETER SETUP
The running times of algorithm (t) is set as 50, the data are
randomly selected to construct different IR datasets 50 times;
and the average results are taken as the final outcome. And
thus, random forest is an ensemble algorithm which has
good performance, so the running parameters are setting in
accordance with tradition. The number of the classifiers (H)
is set as 300 [92], [93].The number of the class is 2, 1 is
used to represent the MIN, and 0 represents the MAJ. Our
algorithm is implemented by C++ and Matlab, the accuracy
of the classification, AUC, F1-score, Recall and the accuracy
of MIN and MAJ are used to analyze the effectiveness. The
results of RF, WRF, and CWsRF are distinguished.
IV. RESULTS
A. AUC, F1-SCORE, RECALL
AUC, F1, Recall is commonly used when the performance
of a classifier needs to be evaluated to select a high pro-
portion of minority instances in the dataset. AUC is bene-
ficial for being independent of class distribution and cost;
Recall is a quality measure of completeness/quantity, which
intuitively reflects the proportion of positive samples that
are correctly identified. F-score (F1) is a harmonic mean of
precision and recall, which can be interpreted as a weighted
average of precision and recall. They can distinguish the per-
formances between classifiers when processing imbalanced
data [94]–[96]. These scores reach the optimum value at 1
VOLUME 6, 2018
and the worst score at 0. As shown in TABLE 6, the AUC,
F1, Recall of CWsRF is higher than those of RF and WRF.
Taking CO as an example, with increasing IR, AUC score of
CWsRF has better performances. When IR = 25%, RF (AUC
score of 0.66, F1 core of 0.76, Recall score of 0.69), WRF
(AUC score of 0.73, F1 core of 0.81, Recall score of 0.75).,
whereas CWsRF (AUC score of 0.83, F1 core of 0.85, Recall
score of 0.88). Although CWsRF got improved, it does not
show a particular advantage. When IR = 20%, AUC score
for CWsRF is 0.17, 0.09 higher than those for RF and WRF;
F1 score is 0.06, 0.02 higher than those for RF and WRF;
Recall score is 0.21, 0.13 higher than those for RF and WRF.
When IR = 15% and 10%, CWsRF has marked advantages,
as AUC and Recall score is approximately 0.30 higher than
those for RF and WRF; F1 score is nearly 0.20 higher than
those for RF and WRF. Therefore, CWsRF has more advan-
tages over RF and WRF in dealing with imbalanced data.
B. ACCURACY OF THE MINORITY AND
MAJORITY SAMPLES
In classification in medical diagnostics, the classes of interest
are often scarce, and therefore, in such an unbalanced situa-
tion, accuracy of the minority samples (ACC_MIN) and accu-
racy of the majority samples (ACC_MAJ) are more important
than the accuracy. Therefore, we observed the changes in
these values, especially in ACC_MIN.
As shown in TABLE 7, when the imbalance increases,
ACC_MAJ, the ability to recognize the majority class,
changes little, whereas ACC_MIN, the ability to recog-
nize the minority class, decreases. However, ACC_MIN of
CWsRF is less affected. ACC_MIN of CWsRF is better than
those of RF and WRF for different IRs. Considering the CO
samples as an example, with the increase in IR, ACC_MIN
of CWsRF shows a distinct advantage. When IR = 25%,
ACC_MIN of RF and WRF are 0.68 and 0.76, respectively,
whereas that of CWsRF is 0.87. When IR = 20%, ACC_MIN
4645
 M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

TABLE 7. ACC_MIN and ACC_MAJ in different IR, where IR = 25%, 20%, 15%, and 10%.

FIGURE 4. 1CWsRF−RF (ACC_MIN improvement (%) between CWsRF and RF)1CWsRF−RF = (ACC_MINCWsRF − ACC_MINRF )/ACC_MINRF .

FIGURE 5. 1CWsRF−WRF (ACC_MIN improvement (%) between CWsRF and WRF) 1CWsRF−WRF = (ACC_MINCWsRF − ACC_MINWRF )/ACC_MINWRF . (a) SPE.
(b) WD. (c) MA. (d) CO. (e) OST.

of RF and WRF are 0.59 and 0.67, respectively, whereas that
of CWsRF is 0.79; when IR = 15% and 10%, ACC_MIN of
RF and WRF, which decreased more obviously, are approxi-
mately 0.40, whereas that of CWsRF is 0.70. With increasing
imbalance, the change in CWsRF is not sensitive to RF and
WRF. Therefore, CWsRF can better identify the minority
class.
Fig. 4 and Fig. 5 show that all the results were posi-
tive, so CWsRF performed better than RF and WRF. Addi-
tionally, with increasing imbalance, especially for IR =
15% and 10%, 1CWsRF−RF and 1CWsRF−WRF are increased,
and thus, CWsRF is more advantageous than RF and WRF.
C. ACC
As shown in Fig. 6, the accuracy of all algorithms is approxi-
mately with 80%-90% due to the increase in imbalance when
the performance is not sufficient to identify the data well, and
the external results are reflected by the increase in accuracy.
For example, when there are 100 data points, 90 of which
belong to the majority class and 10 to the minority class,
if the classifier divides the 100 data points into the majority
class, the correct rate is also 90%. Hence, high accuracy does
not mean good performance, and it is necessary to combine
the classification accuracies of the majority class and the
minority class. An algorithm can be considered a good class

4646 VOLUME 6, 2018

M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

FIGURE 6. Accuracy of different algorithms per IR: (a) IR = 25%, (b) IR = 20%, (c) IR = 15% and (d) IR = 10%.

imbalanced classification algorithm if it fulfills the following Generally, L divides the samples into two classes. If D > 0,
conditions: accuracy without loss (or little loss), increased the samples are classified to MIN, otherwise are MAJ. Larger
AUC, and accuracy of classifying both the minority and distances between the samples and L lead to less misclassifi-
majority samples. Since the accuracy rate is high (greater than cation. Therefore, if DCWsRF − DRF > 0, CWsRF has better
80%) combined with high AUC and ACC_MIN, that CWsRF performances than RF in minority class. The equation is:
achieves better performance.
DCWsRF = (−1) × vtraini,MAJ + (1) × vtraini,MIN + AP
V. DISCUSSION (14)
The Performance and complexity are discussion in the
section. D is calculated from the training set for which is used
to build model; x0 is vtraini,MIN , y0 is vtraini,MIN , C is
A. PERFORMANCE AP.A is −1, B is 1.
The performances of MIN are increased while maintain Substituting (8) into (14), gives:
the accuracy of MAJ (shown in TABLE 6, section IV). XHi,MAJ
So, the performance of MIN has improved significantly, so it DCWsRF = − vi,j,MAJ × wj,MAJ
j=1
is discussed. (Due to limited space, the performance for MAJ XHi,MIN
has not been mentioned.) + vi,jMIN × wj,MIN + AP (15)
j=1
According to the theory of classification, the distance (D) XHi,MAJ XHi,MIN
of the sample to the classification line was used to evaluate DRF =− vi,jMAJ + vi,jMIN (16)
j=1 j=1
the performance of the algorithms. D is calculated from the XHi,MAJ XHi,MIN
DWRF =− vi,j,MAJ × wj + vi,j,MAJ ×wj
measured point to the threshold line. Larger distances lead to j=1 j=1
less misclassification. (17)
There is a sample P (x0 , y0 ), its location is determined by
(X , Y ), X is the number of votes for MAJ, and Y is the number DCWsRF and DRF are compared here, write:
of votes for MIN. The line L is Ax+By+C = 0. The samples XHMAJ
P (x0 , y0 ) can be relatively expressed as D from (x0 , y0 ) to the DCWsRF −DRF = − vi,j,MAJ × wj,MAJ
j=1
line L, Q is the pedal of P on L. Q can be express as XHMIN
+ vi,jMIN × wj,MIN + AP
 2 j=1
B x0 − ABy0 − AC A2 y0 − ABx0 − BC

,
XHMAJ XHMIN
(11) + vi,jMAJ − vi,jMIN
A2 + B2 A2 + B2 j=1 j=1
 2 2
2 B x0 − ABy0 − AC Since Hc is the classifiers which belong to c class, implying
|PQ| = − x0
A2 + B2 that vi,j,MAJ vi,jMIN can replace by 1, lead to:
 2 2
A y0 − ABx0 − BC  XHMAJ XHMAJ
+ − y (12)

0 1− wj,MAJ
A2 + B2 j=1 j=1
Ax0 + By0 + C  XHMIN XHMIN 
D = PQ = √ (13) + wj,MIN − 1 + j (18)
A2 + B2 j=1 j=1

VOLUME 6, 2018 4647

 M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

Since j > 0, Substituting (4), (5) into (18), gives:

PnMIN H
XHMIN i=1 PHj=1 MIN score
i,j,MIN XHMIN XHMAJ
− 1+ 1
j=1 nMIN j=1 j=1
PnMAJ H
XHMAJ i=1 PHj=1 MAJ score
i,j,MAJ
FIGURE 7. The distance of the sample to the classification line.
− (19)
j=1 nMAJ
It can be preceded in two parts:
Part (1): Determine whether the result of (20) is positive:
PnMIN H
XHMIN i=1 PHj=1 MIN score
i,j,MIN XHMIN
− 1
j=1 nMIN j
PnMIN H
XHMIN i=1 PHMIN score
j=1 i,j,MIN
= ( − 1) (20)
j=1 nMIN

PDetermine
nMIN H
whether nMIN greater than
i=1 PHMIN score ,lead to:
j=1 i,j,MIN

XnMIN H XnMIN
PHMIN − 1
i=1 scorei,j,MIN i=1
j=1
XnMIN  H

= PH −1
i=1 MIN
scorei,j,MIN
j=1

Considering the characteristics of classifying imbalanced

FIGURE 8. The result of the equation (24):
dataset, not all of P
them give correctly classification, implying HMIN ∗ H − 1 − HMAJ ∗ (1 − H ).
HMIN
that, H > Max( j=1 scorei,j,MIN ), meaning the result of Ȳ X̄

equation (20) is positive.

Part (2): Determine whether the result of (21) is positive: (23) into (20), (21), gives:
PnMAJ H    
XHi,MAJ z=1 PHj=1 MAJ score H H
z,j,MAJ HMIN ∗ − 1 − HMAJ ∗ 1 − (24)
Hi,MAJ − (21) Ȳ X̄
j=1 nMAJ
Equation (24) expresses our intension on increasing weight
Determine whether nMAJ greater than
P nMAJ H of MIN through HȲ , reducing the weight of MAJ by H .
i=1 PHMAJ , lead to: X̄
j=1 scorei,j,MAJ (For X̄ is larger than Ȳ in the classification of imbalanced data
nX
MAJ   set usually). Since nMAJ nMIN and scorei,j,MIN , scorei,j,MAJ
H are produced RF (according to the procedure 1, section II);
1 − PH
i=1
MAJ
j=1 scorei,j,MAJ therefore, the idea need to analyzed by the datasets instead of
equation of CWsV. The sample CO is took as example.
Considering not all of scorei,j,MAJ are correct, implying From Fig.8, the results are positive; therefore
PHMAJ
that, H > Max( j=1 scorei,j,MAJ ), meaning the result of DCWsRF −DRF is determined to be positive:
equation (21) is negative.
Equation (20) is positive while equation (21) is negative, DCWsRF − DRF > 0
further analysis should carried to determine whether the result The distances of CWsRF are larger than those of RF.
of (20) larger than (21). In order to analyze the problems of Furthermore, the results of the improved distances are shown
easily, write: in Fig. 9. 1CWsV −MV and 1CWsV −WV are positive for all IRs.

H
 Therefore, CWsRF has better generalization ability for MIN
Ȳ = meanHMIN PH (22) than RF and WRF.
MIN
j=1 scorei,j,MIN
X 
nMAJ H B. COMPLEXITY
X̄ = meanHMAJ (23)
i=1
PHMAJ The complexity is O (tHfd · log n) in RF [97], t is the number
j=1 scorei,j,MAJ
of running times, H is the number of the classifiers, f is the
Ȳ is the mean of HMIN classifiers score in MIN, X̄ is the number of the features, n is the number of the data; c is the
mean of HMAJ classifiers score in MAJ. Substituting (22), number of the classes.

4648 VOLUME 6, 2018

M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data
FIGURE 9. Improved distance between the different algorithms per IR in
minority class samples of CO,CWsRF − RF = DCWsRF − DRF ,
CWsRF − WRF = DCWsRF − DWRF where (a), (b), (c), and (d) represent the
improved distances for IR = 25%, 20%, 15%, and 10%, respectively.
One approach is added in WRF compared to RF. It is:
Calculating the weight of each classifier, which mainly
reflected in the number of classifiers, the number of samples.
Its complexity is O (Hn). Therefore, the complexity of WRF
is O (tHfd · log n + tHn)
Two approaches are added in CWsRF compared to RF.
They are (1) the weights of each classifier per class; (2) the
APs based on the number of classifiers.
(1) Approach 1th , which mainly reflected in the num-
ber of classifiers, the number of samples, the complexity
of the number of classes. So the complexity is O (Hnc);
(2) Approach 2th , which has an application of the sorting
process, selecting the maximum AUC value, which mainly
reflected in the number of classifiers. So the complexity is
O (H · log H ).
Therefore, the complexity of CWsRF is O (t · (Hfd · log n+
Hnc + H · log H )). The values (n, k and t) have influence on
the efficiency of CWsRF.
In addition, to evaluate the algorithms’ complexity,
the average running time on the five different datasets were
calculated and the results are shown in TABLE 8. The exper-
iments are conducted on PC ( Intel Core i7-3537U, 2.5 GHz
CPU and 4 GB memory).
TABLE 8. The time cost of the algorithms (seconds).
From the TABLE 8, it shows similar runtime on the dif-
ferent data for the same algorithm. The shortest cost time
of CWsRF is 2.20s (on the dataset with 962 instances,
5 attributes), while the longest is 2.41s (on the dataset with
313 instances, 85 attributes). The results are acceptable in the
practice of medical diagnosis and big complex data will be
tested in the further study.
VOLUME 6, 2018
VI. CONCLUSION
The classification of class imbalanced data is a new research
topic and represents an urgent problem to be solved.
An algorithm (CWsRF) to develop class weights for pro-
cessing imbalanced medical data was proposed. In the study,
the empirical error is taken as the measurement to obtain
class weights of the classifier. The algorithm yields superior
performance than other schemes. The proposed algorithm had
very high accuracy classifying, AUC, F1 and Recall.
This paper is an attempt to improve the ensemble learning
algorithm of RF dealing with binary classification and could
be extended to ensemble learning with other algorithms and
multi-classification problems.
REFERENCES
[1] Y. Zhu and J. Fang, ‘‘Logistic regression-based trichotomous classification
tree and its application in medical diagnosis,’’ Med. Decision Making,
vol. 36, no. 8, pp. 973–989, 2016.
[2] S. D. Zhao, ‘‘Integrative genetic risk prediction using non-parametric
empirical Bayes classification,’’ Biometrics, vol. 73, no. 2, pp. 582–592,
2017.
[3] E. Dong, C. Li, L. Li, S. Du, A. N. Belkacem, and C. Chen, ‘‘Classification
of multi-class motor imagery with a novel hierarchical SVM algorithm
for brain-computer interfaces,’’ Med. Biol. Eng. Comput., vol. 55, no. 10,
pp. 1809–1818, 2017.
[4] M. Zhu, J. Xia, M. Yan, G. Cai, J. Yan, and G. Ning, ‘‘Dimension-
ality reduction in complex medical data: Improved self-adaptive niche
genetic algorithm,’’ Comput. Math. Methods Med., vol. 2015, Oct. 2015,
Art. no. 794586, doi: 10.1155/2015/794586.
[5] M. Durgadevi and R. Kalpana, ‘‘Medical distress prediction based on
classification rule discovery using ant-miner algorithm,’’ in Proc. 11th Int.
Conf. Intell. Syst. Control (ISCO), Jan. 2017, pp. 88–92.
[6] M. Chen, Y. Hao, K. Hwang, L. Wang, and L. Wang, ‘‘Disease prediction
by machine learning over big data from healthcare communities,’’ IEEE
Access, vol. 5, no. 1, pp. 8869–8879, 2017.
[7] M. Chen, X. Shi, Y. Zhang, D. Wu, and M. Guizani, ‘‘Deep
features learning for medical image analysis with convolutional
autoencoder neural network,’’ IEEE Trans. Big Data, 2017,
doi: 10.1109/TBDATA.2017.2717439.
[8] B. A. Bak and J. L. Jensen, ‘‘High dimensional classifiers in the imbalanced
case,’’ Comput. Stat. Data Anal., vol. 98, pp. 46–59, Jun. 2016.
[9] Y. Zhang, P. Fu, W. Liu, and G. Chen ‘‘Imbalanced data classification based
on scaling kernel-based support vector machine,’’ Neural Comput. Appl.,
vol. 25, nos. 3–4, pp. 927–935, 2014.
[10] C. K. Maurya, D. Toshniwal, and G. V. Venkoparao, ‘‘Online sparse class
imbalance learning on big data,’’ Neurocomputing, vol. 216, pp. 250–260,
Dec. 2016.
[11] S. Al-Stouhi and C. K. Reddy, ‘‘Transfer learning for class imbal-
ance problems with inadequate data,’’ Knowl. Inf. Syst., vol. 48, no. 1,
pp. 201–228, 2016.
[12] M. El-Banna, ‘‘Modified Mahalanobis Taguchi system for imbalance
data classification,’’ Comput. Intell. Neurosci., vol. 2017, Jul. 2017,
Art. no. 5874896.
[13] B. Mirza et al., ‘‘Efficient representation learning for high-dimensional
imbalance data,’’ in Proc. IEEE Int. Conf. Digit. Signal Process. (DSP),
Oct. 2016, pp. 511–515.
[14] M. M. Al-Rifaie and H. A. Alhakbani, ‘‘Handling class imbalance in direct
marketing dataset using a hybrid data and algorithmic level solutions,’’ in
Proc. SAI Comput. Conf. (SAI), Jul. 2016, pp. 446–451.
[15] H. Lu, K. Yang, and J. Shi, ‘‘Constraining the water imbalance in a
land data assimilation system through a recursive assimilation scheme,’’
in Proc. IEEE Int. Geosci. Remote Sens. Symp. (IGARSS), Jul. 2016,
pp. 2993–2996.
[16] S. Pouyanfar and S.-C. Chen, ‘‘Automatic video event detection for
imbalance data using enhanced ensemble deep learning,’’ Int. J. Semantic
Comput., vol. 11, no. 1, pp. 85–109, 2017.
[17] W. Mao, J. Wang, and Z. Xue, ‘‘An ELM-based model with sparse-
weighting strategy for sequential data imbalance problem,’’ Int. J. Mach.
Learn. Cybern., vol. 8, no. 4, pp. 1333–1345, 2017.
4649
 M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data
[18] J. Zhai, S. Zhang, and C. Wang, ‘‘The classification of imbalanced large
data sets based on MapReduce and ensemble of ELM classifiers,’’ Int.
J. Mach. Learn. Cybern., vol. 8, no. 3, pp. 1009–1017, 2017.
[19] A. Amin et al., ‘‘Comparing oversampling techniques to handle the
class imbalance problem: A customer churn prediction case study,’’ IEEE
Access, vol. 4, pp. 7940–7957, 2016.
[20] K. Jiang, J. Lu, and K. Xia, ‘‘A novel algorithm for imbalance data
classification based on genetic algorithm improved SMOTE,’’ Arabian
J. Sci. Eng., vol. 41, no. 8, pp. 3255–3266, 2016.
[21] X. Zhang, Q. Song, G. Wang, K. Zhang, L. He, and X. Jia, ‘‘A dissimilarity-
based imbalance data classification algorithm,’’ Appl. Intell., vol. 42, no. 3,
pp. 544–565, 2015.
[22] J. Wang, J. Z. Feng, and Z. Han, ‘‘Discriminative feature selection based
on imbalance SVDD for fault detection of semiconductor manufacturing
processes,’’ J. Circuits, Syst. Comput., vol. 25, no. 11, p. 1650143, 2016.
[23] P. Vorraboot, S. Rasmequan, K. Chinnasarn, and C. Lursinsap, ‘‘Improving
classification rate constrained to imbalanced data between overlapped and
non-overlapped regions by hybrid algorithms,’’ Neurocomputing, vol. 152,
pp. 429–443, Mar. 2015.
[24] P. Du, A. Samat, B. Waske, S. Liu, and Z. Li, ‘‘Random forest and rotation
forest for fully polarized SAR image classification using polarimetric
and spatial features,’’ ISPRS J. Photogramm. Remote Sens., vol. 105,
pp. 38–53, Jul. 2015.
[25] M. Belgiu and L. Drǎguţ, ‘‘Random forest in remote sensing: A review of
applications and future directions,’’ ISPRS J. Photogramm. Remote Sens.,
vol. 114, pp. 24–31, Apr. 2016.
[26] L.-I. Tong, K.-H. Chang, P.-Y. Wu, and Y.-C. Chan, ‘‘Using dual response
surface methodology as a benchmark to process multi-class imbalanced
data,’’ J. Ind. Prod. Eng., vol. 34, no. 2, pp. 147–158, 2017.
[27] H. Lee, E. Kim, and S. Kim, ‘‘Anomalous propagation echo classification
of imbalanced radar data with support vector machine,’’ Adv. Meteorol.,
vol. 2016, pp. 1–13, Jan. 2016.
[28] M. J. Fernández-Gómez, G. Asencio-Cortés, A. Troncoso, and
F. Martínez-Álvarez, ‘‘Large earthquake magnitude prediction in
Chile with imbalanced classifiers and ensemble learning,’’ Appl. Sci.,
vol. 7, no. 6, p. 625, 2017.
[29] J. Jia, Z. Liu, X. Xiao, B. Liu, and K. C. Chou, ‘‘iPPBS-Opt: A sequence-
based ensemble classifier for identifying protein-protein binding sites by
optimizing imbalanced training datasets,’’ Molecules, vol. 21, no. 1, p. 95,
2016.
[30] U. R. Salunkhe and S. N. Mali, ‘‘Classifier ensemble design for imbalanced
data classification: A hybrid approach,’’ Procedia Comput. Sci., vol. 85,
pp. 725–732, May 2016.
[31] J.-H. Xue and P. Hall, ‘‘Why does rebalancing class-unbalanced data
improve AUC for linear discriminant analysis?’’ IEEE Trans. Pattern Anal.
Mach. Intell., vol. 37, no. 5, pp. 1109–1112, May 2015.
[32] M. Thakong, S. Phimoltares, S. Jaiyen, and C. Lursinsap, ‘‘Fast learn-
ing and testing for imbalanced multi-class changes in streaming data by
dynamic multi-stratum network,’’ IEEE Access, vol. 5, pp. 10633–10648,
2017.
[33] B. Wang and J. Pineau, ‘‘Online bagging and boosting for imbalanced data
streams,’’ IEEE Trans. Knowl. Data Eng., vol. 28, no. 12, pp. 3353–3366,
Dec. 2016.
[34] M. N. Haque, N. Noman, R. Berretta, and P. Moscato, ‘‘Heterogeneous
ensemble combination search using genetic algorithm for class imbalanced
data classification,’’ PLoS ONE, vol. 11, no. 1, p. e0146116, 2016.
[35] P. Blonda, C. Tarantino, A. D’Addabbo, G. Satalino, and G. Pasquariello,
‘‘Combination of multiple classifiers by fuzzy integrals: An application to
synthetic aperture radar (SAR) data,’’ in Proc. IEEE Int. Fuzzy Syst. Conf.,
vol. 3. Dec. 2001, pp. 944–947.
[36] Q. Wang et al., ‘‘A novel ensemble method for imbalanced data learn-
ing: Bagging of extrapolation-SMOTE SVM,’’ Comput. Intell. Neurosci.,
vol. 2017, pp. 1–11, Jan. 2017.
[37] H. Li, Y. Wang, H. Wang, and B. Zhou, ‘‘Multi-window based ensemble
learning for classification of imbalanced streaming data,’’ World Wide Web,
vol. 20, no. 6, pp. 1507–1525, 2017.
[38] S. Ali, A. Majid, S. G. Javed, and M. Sattar, ‘‘Can-CSC-GBE: Developing
Cost-sensitive Classifier with Gentleboost Ensemble for breast cancer
classification using protein amino acids and imbalanced data,’’ Comput.
Biol. Med., vol. 73, pp. 38–46, Jun. 2016.
[39] J. Kevric, S. Jukic, and A. Subasi, ‘‘An effective combining classifier
approach using tree algorithms for network intrusion detection,’’ Neural
Comput. Appl., vol. 28, pp. 1051–1058, Dec. 2016.
4650
[40] H. Noçairi, C. Gomes, M. Thomas, and G. Saporta, ‘‘Improving stacking
methodology for combining classifiers; applications to cosmetic industry,’’
Electron. J. Appl. Stat. Anal., vol. 9, no. 2, pp. 340–361, 2016.
[41] T.-M. Chan, Y. Li, C.-C. Chiau, J. Zhu, J. Jiang, and Y. Huo, ‘‘Imbalanced
target prediction with pattern discovery on clinical data repositories,’’ BMC
Med. Inform. Decision Making, vol. 17, p. 47, Apr. 2017.
[42] Y. Zhang, J. Ren, and J. Jiang, ‘‘Combining MLC and SVM classifiers
for learning based decision making: Analysis and evaluations,’’ Comput.
Intell. Neurosci., vol. 2015, pp. 1–8, May 2015.
[43] D. Zhou, Y. Tang, and W. Jiang, ‘‘A modified belief entropy in
Dempster–Shafer framework,’’ PLoS ONE, vol. 12, no. 5, p. e0176832,
2017.
[44] X. Deng, W. Jiang, and J. Zhang, ‘‘Zero-Sum matrix game with payoffs
of Dempster–Shafer belief structures and its applications on sensors,’’
Sensors, vol. 17, no. 4, p. 922, 2017.
[45] W. Chen, H. R. Pourghasemi, and Z. Zhao, ‘‘A GIS-based comparative
study of Dempster–Shafer, logistic regression and artificial neural network
models for landslide susceptibility mapping,’’ Geocarto Int., vol. 32, no. 4,
pp. 367–385, 2017.
[46] A. M. Al-Abadi, ‘‘The application of Dempster–Shafer theory of evidence
for assessing groundwater vulnerability at Galal Badra basin, Wasit gover-
norate, east of iraq,’’ Appl. Water Sci., vol. 7, no. 4, pp. 1725–1740, 2017.
[47] J. Wang, K. Qiao, Z. Zhang, and F. Xiang, ‘‘A new conflict management
method in Dempster–Shafer theory,’’ Int. J. Distrib. Sensor Netw., vol. 13,
no. 3, pp. 1–11, 2017.
[48] C.-D. Zheng, Y. Zhang, and Z. Wang, ‘‘Novel stability condition of
stochastic fuzzy neural networks with Markovian jumping under impulsive
perturbations,’’ Int. J. Mach. Learn. Cybern., vol. 7, no. 5, pp. 795–803,
2016.
[49] P. Chen and D. Zhang, ‘‘Constructing support vector machines ensemble
classification method for imbalanced datasets based on fuzzy integral,’’
in Modern Advances in Applied Intelligence (Lecture Notes in Computer
Science), vol. 7. Berlin, Germany: Springer, 2014, pp. 70–76.
[50] R. Tang, Y. Zhu, and G. Chen, ‘‘Imbalanced data classification method
based on clustering and voting mechanism,’’ in Proc. Int. Conf. Inf., 2013,
pp. 667–674.
[51] R. Hidayati, K. Kanamori, L. Feng, and H. Ohwada, ‘‘Implementing major-
ity voting rule to classify corporate value based on environmental efforts,’’
in Data Mining and Big Data (Lecture Notes in Computer Science),
vol. 211. Berlin, Germany: Springer, 2016, pp. 59–66.
[52] M. C. Çolaka, C. Çolakb, N. Erdila, and A. K. Arslan, ‘‘Investigating
optimal number of cross validation on the prediction of postoperative atrial
fibrillation by voting ensemble strategy,’’ Turkiye Klinikleri J. Biostat.,
vol. 8, no. 1, pp. 30–35, 2016.
[53] A. Tamvakis, G. E. Tsekouras, A. Rigos, C. Kalloniatis,
C. N. Anagnostopoulos, and G. Anastassopoulos, ‘‘A methodology to
carry out voting classification tasks using a particle swarm optimization-
based neuro-fuzzy competitive learning network,’’ Evolving Syst., vol. 8,
no. 1, pp. 49–69, 2017.
[54] S. Abbasi, A. Shahriari, and Y. Nemati, ‘‘Retracted: A novel voting math-
ematical rule classification for image recognition,’’ in Computational Sci-
ence and Its Applications—ICCSA (Lecture Notes in Computer Science),
vol. 8. Berlin, Germany: Springer, 2016, pp. 257–270.
[55] T. Subbulakshmi and R. R. Raja, ‘‘An ensemble approach for sentiment
classification: Voting for classes and against them,’’ ICTACT J. Soft
Comput., vol. 6, no. 4, pp. 1281–1286, 2016.
[56] B. Xia, H. Jiang, H. Liu, and D. Yi, ‘‘A novel hepatocellular carcinoma
image classification method based on voting ranking random forests,’’
Comput. Math. Methods Med., vol. 2016, Apr. 2016, Art. no. 2628463.
[57] A. Linden and P. R. Yarnold, ‘‘Using classification tree analysis to generate
propensity score weights,’’ J. Eval. Clin. Pract., vol. 23, no. 4, pp. 703–712,
2017.
[58] C. De Stefano, F. Fontanella, and A. S. di Freca, ‘‘A novel naive Bayes
voting strategy for combining classifiers,’’ in Proc. Int. Conf. Frontiers
Handwriting Recognit., Sep. 2012, pp. 467–472.
[59] G. Rogova, ‘‘Combining the results of several neural network classifiers,’’
in Classic Works of the Dempster-Shafer Theory of Belief Functions,
vol. 219. Manchester, U.K.: IEEE, 2008, pp. 683–692.
[60] S. R. Kheradpisheh, A. Nowzari-Dalini, R. Ebrahimpour, and
M. Ganjtabesh, ‘‘An evidence-based combining classifier for brain
signal analysis,’’ PLoS ONE, vol. 9, no. 1, p. e84341, 2014.
[61] Y. Bi, D. Bell, H. Wang, G. Guo, and K. Greer, ‘‘Combining multiple
classifiers using Dempster’s rule of combination for text categorization,’’
in Proc. Int. Conf. Modeling Decisions Artif. Intell., 2004, pp. 127–138.
VOLUME 6, 2018
M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data
[62] S. Chandana, H. Leung, and K. Trpkov, ‘‘Staging of prostate cancer
using automatic feature selection, sampling and Dempster–Shafer fusion,’’
Cancer Inform., vol. 2009, no. 7, pp. 57–73, Feb. 2009.
[63] Y. Li and J. Jingping, ‘‘New algorithm for combining classifiers based
on fuzzy integral and genetic algorithms,’’ Proc. SPIE, vol. 4554,
pp. 176–181, Sep. 2001.
[64] J. Svec and J. Hamilton, ‘‘Endogenous voting weights for elected represen-
tatives and redistricting,’’ Constitutional Political Economy, vol. 26, no. 4,
pp. 434–441, 2015.
[65] Q. Wu, Y. Ye, Y. Liu, and M. K. Ng, ‘‘SNP selection and classification of
genome-wide SNP data using stratified sampling random forests,’’ IEEE
Trans. Nanobiosci., vol. 11, no. 3, pp. 216–226, Sep. 2012.
[66] Y. Ye, Q. Wu, J. Z. Huang, M. K. Ng, and X. Li, ‘‘Stratified sampling for
feature subspace selection in random forests for high dimensional data,’’
Pattern Recognit., vol. 46, no. 3, pp. 769–787, 2013.
[67] J. Sun, G. Zhong, J. Dong, H. Saeeda, and Q. Zhang, ‘‘Cooperative profit
random forests with application in ocean front recognition,’’ IEEE Access,
vol. 5, pp. 1398–1408, 2017.
[68] W. Lin, Z. Wu, L. Lin, A. Wen, and J. Li, ‘‘An ensemble random
forest algorithm for insurance big data analysis,’’ IEEE Access, vol. 5,
pp. 16568–16575, 2017.
[69] L. Breiman, ‘‘Random forests,’’ Mach. Learn., vol. 45, no. 1, pp. 5–32,
2001.
[70] M. E. H. Daho, N. Settouti, M. E. A. Lazouni, and M. E. A. Chikh,
‘‘Weighted vote for trees aggregation in random forest,’’ in Proc. Int. Conf.
Multimedia Comput. Syst. (ICMCS), Apr. 2014, pp. 438–443.
[71] T. Perry and M. Bader-El-Den, ‘‘Imbalanced classification using geneti-
cally optimized random forests,’’ in Proc. Companion Publication Annu.
Conf. Gen. Evol. Comput., 2015, vol. 15. no. 7, pp. 1453–1454.
[72] C. A. Ronao and S.-B. Cho, ‘‘Random forests with weighted voting for
anomalous query access detection in relational databases,’’ in Artificial
Intelligence and Soft Computing (Lecture Notes in Computer Science),
vol. 2015. New York, NY, USA: ACM, 2015, pp. 36–48.
[73] S. A. Naghibi, K. Ahmadi, and A. Daneshi, ‘‘Application of support
vector machine, random forest, and genetic algorithm optimized random
forest models in groundwater potential mapping,’’ Water Resour. Manage.,
vol. 31, no. 9, pp. 2761–2775, 2017.
[74] A. M. Youssef, H. R. Pourghasemi, Z. S. Pourtaghi, and M. M. Al-Katheeri,
‘‘Landslide susceptibility mapping using random forest, boosted regression
tree, classification and regression tree, and general linear models and
comparison of their performance at Wadi Tayyah Basin, Asir Region, Saudi
Arabia,’’ Landslides, vol. 13, no. 5, pp. 839–856, 2016.
[75] S. A. Naghibi, H. R. Pourghasemi, and B. Dixon, ‘‘GIS-based groundwater
potential mapping using boosted regression tree, classification and regres-
sion tree, and random forest machine learning models in iran,’’ Environ.
Monitor. Assessment, vol. 188, no. 1, p. 44, 2016.
[76] L. I. Kuncheva and J. J. Rodríguez, ‘‘A weighted voting framework for
classifiers ensembles,’’ Knowl. Inf. Syst., vol. 38, no. 2, pp. 259–275, 2014.
[77] Y. Bachrach, Y. Filmus, J. Oren, and Y. Zick, ‘‘Analyzing power in
weighted voting games with super-increasing weights,’’ in Proc. Int. Symp.
Algorithmic Game Theory, 2012, pp. 169–181.
[78] T. Hayes, S. Usami, R. Jacobucci, and J. J. McArdle, ‘‘Using clas-
sification and regression trees (CART) and random forests to analyze
attrition: Results from two simulations,’’ Psychol. Aging, vol. 30, no. 4,
pp. 911–929, 2015.
[79] H. Ishwaran, U. B. Kogalur, E. H. Blackstone, and H. Lauer, ‘‘Random
survival forests,’’ Ann. Appl. Stat., vol. 2, no. 3, pp. 841–860, 2008.
[80] N. Tóth and B. Pataki, ‘‘Classification confidence weighted majority voting
using decision tree classifiers,’’ Int. J. Intell. Comput. Cybern., vol. 1, no. 2,
pp. 169–192, 2008.
[81] A. F. R. Rahman and M. C. Fairhurst, ‘‘Multiple classifier decision com-
bination strategies for character recognition: A review,’’ Document Anal.
Recognit., vol. 5, no. 4, pp. 166–194, 2003.
[82] A. Arnaiz-González, J. F. Díez-Pastor, C. García-Osorio, and
J. J. Rodríguez, ‘‘Random feature weights for regression trees,’’ Progr.
Artif. Intell., vol. 5, no. 2, pp. 91–103, 2016.
[83] S. J. Winham, R. R. Freimuth, and J. M. Biernacka, ‘‘A weighted random
forests approach to improve predictive performance,’’ Stat. Anal. Data
Mining, vol. 6, no. 6, pp. 496–505, 2013.
[84] M. Das and S. Bhattacharya, ‘‘A modified history based weighted average
voting with soft-dynamic threshold,’’ in Proc. Int. Conf. Adv. Comput. Eng.,
Jun. 2010, pp. 217–222.
VOLUME 6, 2018
[85] C. Tang, C. Hou, P. Wang, and Z. Song, ‘‘Salient object detection using
color spatial distribution and minimum spanning tree weight,’’ Multimed
Tools Appl., vol. 75, no. 12, pp. 6963–6978, 2016.
[86] J. Chiquet, G. Rigaill and P. Gutierrez, ‘‘Fast tree inference with weighted
fusion penalties,’’ J. Comput. Graph. Stat., vol. 26, no. 1, pp. 205–216,
2017.
[87] Z. Xu, C. Voichita, S. Drǎghici, and R. Romero, ‘‘Z-bag: A classification
ensemble system with posterior probabilistic outputs,’’ Comput. Intell.,
vol. 29, no. 2, pp. 310–330, 2013.
[88] C. E. DeSantis, F. Bray, J. Ferlay, J. Lortet-Tieulent, B. O. Anderson,
and A. Jemal, ‘‘International variation in female breast cancer incidence
and mortality rates,’’ Cancer Epidemiol. Biomarkers Prevention, vol. 24,
no. 10, pp. 1495–1506, 2015.
[89] O. Johnell and J. Kanis, ‘‘Epidemiology of osteoporotic fractures,’’ Osteo-
porosis Int., vol. 16, pp. S3–S7, Mar. 2005.
[90] S. Janitza, C. Strobl, and A.-L. Boulesteix, ‘‘An AUC-based permutation
variable importance measure for random forests,’’ BMC Bioinf., vol. 14,
pp. 119–130, Apr. 2013.
[91] A. I. Marqués, V. García, and J. S. Sánchez, ‘‘On the suitability of
resampling techniques for the class imbalance problem in credit scoring,’’
J. Oper. Res. Soc., vol. 64, no. 13, pp. 1060–1070, 2013.
[92] A. Cuzzocrea, S. L. Francis, and M. M. Gaber, ‘‘An information-theoretic
approach for setting the optimal number of decision trees in random
forests,’’ in Proc. IEEE Int. Conf. Syst., Man, Cybern., vol. 177. Oct. 2013,
pp. 1013–1019.
[93] P. Latinne, O. Debeir, and C. Decaestecker, ‘‘Limiting the number of trees
in random forests,’’ in Multiple Classifier Systems, vol. 2013. Manchester,
U.K.: IEEE, 2001, pp. 178–187.
[94] R. K. Shahzad, M. Fatima, N. Lavesson, and M. Boldt, ‘‘Consensus deci-
sion making in random forests,’’ in Machine Learning, Optimization, and
Big Data (Lecture Notes in Computer Science). Berlin, Germany: Springer
2015, pp. 347–358.
[95] J. Hu, ‘‘Automated detection of driver fatigue based on AdaBoost classifier
with EEG signals,’’ Frontiers Comput. Neurosci., vol. 11, no. 72, pp. 1–10,
2017.
[96] J. Hu, ‘‘Automated detection of driver fatigue based on AdaBoost classifier
with EEG signals,’’ Frontiers Comput. Neurosci., vol. 11, no. 8, pp. 1–10,
2017.
[97] G. Biau, ‘‘Analysis of a random forests model,’’ J. Mach. Learn. Res.,
vol. 13, pp. 1063–1095, Apr. 2012.
MIN ZHU received the M.S. degree from the
College of Computer Science and Technology,
Guizhou University, Guiyang, China, in 2006. She
is currently working toward the Ph.D. degree at
Zhejiang University, Hangzhou, China. She is cur-
rently a Senior Engineer at the Guizhou Key Lab-
oratory of Agricultural Bioengineering, Guizhou
University. Her research interests include data
mining, pattern recognition, and classification.
JING XIA received the B.S. degree in biomedi-
cal engineering from Zhejiang University, China,
in 2013. She is currently working toward the Ph.D.
degree in biomedical engineering at Zhejiang Uni-
versity, China. Her research interests focus on
intelligent medical diagnosis.
4651
 M. Zhu et al.: Class Weights Random Forest Algorithm for Processing Class Imbalanced Medical Data

XIAOQING JIN received the Ph.D. degree from
the Department of Acupuncture, Zhejiang Chinese
Medical University, China. She is currently the
Head of the Department of Acupuncture, Zhejiang
Hospital, China.
JING YAN received the M.S. degree from the
Department of Cardiology, Zhejiang University,
China. He is currently the Dean of Zhejiang
Hospital, China.

MOLEI YAN received the Ph.D. degree from

the Department of Internal Medicine, Zhejiang
Chinese Medical University, China. She is cur-
rently a Physician at the Department of ICU, Zhe-
jiang Hospital, China.

GUOLONG CAI received the M.S. degree from
the Department of Cardiology, Zhejiang Univer-
sity, China. He is currently a Physician at the
Department of ICU, Zhejiang Hospital, China.
GANGMIN NING received the Dr.-Ing. degree
from the Deparment of Biomedical Engineering,
TU Ilmenau, Germany. He is a currently a Profes-
sor at the Department of Biomedical Engineering,
Zhejiang University, China.

4652 VOLUME 6, 2018