2023 International Conference on Integrated Intelligence and Communication Systems (ICIICS)

Application of ID3 Decision Tree Classification

Algorithm in Mathematical data Analysis
2023 International Conference on Integrated Intelligence and Communication Systems (ICIICS) | 979-8-3503-1545-5/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICIICS59993.2023.10421137

Ping Zang
Shandong Transport Vocational College
Weifang Shandong, China
[email protected]

Abstract—ID3 decision tree is the most extensive method. It development of ID3 algorithm at home and abroad started late,
can be used to classify, identify and predict data information, it is widely used in various fields because of its good
and the whole system can be divided into several stages performance. For the related problems involved in data
according to its nature. It plays an important role in mining, such as knowledge discovery and utilization, decision
mathematical data analysis. Therefore, based on ID3 decision analysis and so on, it has been widely valued and widely used,
classification algorithm, this paper studies mathematical data and a series of research results have been obtained. For
analysis. This paper firstly introduces the decision tree example, after studying the information retrieval based on
classification algorithm, and paper expounds the ID3 decision web-s structure, the professor of Massachusetts Institute of
tree classification algorithm in data analysis, the application of
technology proposed an information search system that
mathematics algorithm in the application of mathematical data
analysis process is introduced, and the performance of ID3
combines multiple features and algorithms to realize the
decision tree classification algorithm was tested. Finally, the test integrated search function. Another American scholar
results show that for a high-resolution decision tree, if the proposed a "rank integral" algorithm based on fuzzy set
traditional method is used for classification, it will increase the theory [5-6]. A lot of research has been done in this field in
amount of calculation and time. When the ID3 decision number China. Some scholars obtained through the analysis of the
algorithm is used to improve the model, it can effectively original sample data. When the value range of a parameter is
improve the decision accuracy. The mathematical data analysis small, it can be considered that the algorithm has good
time based on the ID3 decision number algorithm increases performance. Some scholars proposed to combine data
with the increase of the number of data. However, the overall structure with decision tree theory to establish mathematical
time is within 1-4 seconds, and at the same time, the model. It is considered that using this model can improve the
classification accuracy is more than 93%, which shows that the performance indexes such as the ability to solve the attribute
performance of the algorithm is good. value of the global optimal solution problem of the
relationship between the data set and the decision tree method
Keywords—classification algorithm, data analysis, ID3 and the convergence of the algorithm [7-8]. The above
decision tree, mathematical data, model improvement research has laid a research foundation for this paper.
I. INTRODUCTION With the rapid development of science and technology,
With the development of modern industrial technology, computer technology has also been improved to a great extent,
ID3 system plays an important role in the production process. and its application fields are becoming more and more
However, the traditional controller is composed of classical extensive. As one of the important methods in mathematical
algorithms and reasoning theory. This makes the performance analysis, ID3 algorithm is widely used in all aspects. This
of the control system reduced a lot or even no efficient paper mainly expounds, discusses and summarizes the
operation, seriously affecting the control design and research direction. Firstly, this paper introduces the relevant
economic benefits; At the same time, due to the limited knowledge of data mining, ID3 system and decision tree
capacity of data memory, complex program and other reasons, classification, then combines the two kinds of machine
the algorithm cannot be used in the online real-time analysis learning methods and applies them to specific problems
process, resulting in a large number of errors or delayed according to the actual situation, puts forward some opinions
output results. ID3 programming is a very important part of on this subject, and gives the corresponding result evaluation.
mathematics, also known as computer model. In real life, we Section 2 focuses on the related literature. Section 3
often encounter problems such as data collection, storage and introduces the proposed methods. Section 4 presents
management. This requires sorting out, summarizing and assessments of the presented approach and analyses the
analyzing these large amounts of information to help results. Section 5 discussed the analysis of proposed method.
decision-makers make correct decisions and adopt reasonable Section 6 concludes the study and makes recommendations
and effective strategies to solve this problem. For practical for further research.
problems, the value range of many uncertain parameters will
affect decision makers to make correct choices [1-2]. II. RELATED WORKS
Therefore, we need an algorithm to calculate whether each Fatima Es-Sabery et al. [9] developed the MapReduce
objective function corresponding to the constraint has opinion mining for Covid-19 related tweets classification
changed under different variables and the possible results of using enhanced ID3 decision tree classifier. The several
such changes. This requires that the computer must be able to feature extractors were utilized to effectively detect and
adjust its output data set in real time according to the specific capture the relevant data from the given tweets which includes
situation [3-4]. bag-of-words, word embedding, FastText. A multiple feature
Many scholars have studied ID3 decision tree selector was applied to reduce the high feature’s
classification algorithm. Although the research and dimensionality which includes, Gini Index, Chi-square,

979-8-3503-1545-5/23/$31.00 ©2023 IEEE

orized licensed use limited to: National Hellenic Research Foundation & National Documentation Centre. Downloaded on September 25,2024 at 13:11:13 UTC from IEEE Xplore. Restrictions ap
 Information Gain and Gain Ratio. At last, the attained features III. ID3 DECISION TREE CLASSIFICATION ALGORITHM
were utilized to carry out the classification task by utilizing
an enhanced ID3 classifier. Thence the empirical result A. Decision Tree Classification Algorithm
proved that the information gain is the most effective feature 1) Definition: The decision tree classification based on
selector. However, the developed method was integrate with decision variables and data is a set of characteristic vector of
fuzzy logic theory to handle the continuous-valued features the learning process. By analyzing a large number of attribute
which takes into consideration of various parameters in values, the mathematical model is established. This
feature extractors and selectors. algorithm adopts many traditional methods. It can be applied
Arun Agarwal et al. [10] suggested the mathematical to machine recognition, statistical reasoning and pattern
model based on modified ID3 algorithm for healthcare recognition. Each sample set contains a certain number of
diagnostics model. This research proposes a decision-support representative records that can represent the whole sample
framework for any disease prediction in the healthcare sector. object, with a large number of representations and a small
The Modified ID3 algorithm is implemented in python number of discontinuous points. From this data set, a variable
programming by using a reduced feature set of the Hepatitis is randomly selected as the eigenvector quantum set of
C virus dataset. The essential features were identified using
the Chi-square method, and further, we have tested the health classification attribute values. Decision tree classification is
care diagnostic proposed model with ID3, RF method, SVM a way to classify multiple data samples according to
in the HCV dataset. The Mod-ID3 approach improves the F1 attributes, thus forming a network model composed of many
score while maintaining an acceptable recall value. However, subclasses. It can use a feature point to describe these
the developed method was required to expand over real characteristics. Branch out from the node and obtain the Bi
healthcare datasets for predicting diseases accurately. conclusion tree from the leaf node of the decision. One of the
Yi Liu and Shuo Yang [11] presented the applications of biggest advantages of machine learning algorithm based on
decision tree-based classification algorithm on content decision tree is that users do not need to master many
marketing. A decision tree-based methods are introduced to knowledge in the process of practice, and can only use an
the field of content marketing, and decision tree-based attribute description for the example of practice [14-15]. It
methods intrinsically follow the process of human decision uses mathematical statistical methods and statistical
making. Specifically, this paper considers a well-known principles to construct a simple rule structure based on the
method, called C4.5, which can deal well with continuous decision table. In practical application, theoretical prediction,
values. Based on four validation metrics, experimental results method determination, optimal solution selection,
obtained from several machine learning-based methods comprehensive optimization and other problems all need a
indicate that the C4.5-based decision tree method has the
person who can accurately reflect the essential characteristics
ability to handle the content marketing dataset. Besides, it
cannot learn the interaction among features, which highly of things in the real world, has certain knowledge and
limits its experimental performances. experience (such as mathematical chemistry theory), and can
recognize the laws of the objective world and make scientific
Li Yu et al. [12] implemented the online data migration judgment for people. Therefore, when making prediction,
model and ID3 algorithm in sports competition action data We must first define a sample as the relationship between the
mining application. This research proposes a k -nearest
input point and other variables (i.e. "class") to describe all
neighbor-based ID3 optimization algorithm, which selects
values similar to k-nearest neighbors to fill in the missing possible results on these data sets. A certain type has a certain
values for the attribute missing problem of the traditional ID3 degree of similarity or no difference. Each group has a value
algorithm. Based on this, the improved algorithm is applied greater than 0 or less than 1 in different attribute sets, which
to the online data migration of sports competition actions, and is called the minimum classification group [16-17].
the application effect is evaluated. The clustering results have 2) ID3 decision tree classification algorithm: ID3
some errors and are only provided as a reference to the also- algorithm is a mathematical model for classifying data sets.
ran scholars for reference and cannot replace the actual also- It can calculate its corresponding parameters by calculating
ran judgments. the relationship between different point sets in ID sample
Sirichanya Chanmee and Kraisak Kesorn [13] developed space. According to the value of linear function of each
the semantic decision trees for ID3-based algorithm using sample, it can be divided into 1, 2 and 4 categories. Each data
knowledge base. An ontology summarization technique was object has its corresponding eigenvector. The most
adopted to fine-tune the splitting condition of the DT. This significant advantages of ID3 algorithm are: the structure of
adjustment is influenced by the attribute’s importance value, the algorithm itself is simple and clear, convenient and easy
extracted directly from the ontology. This process leads to the to use, and the machine learning ability is strong. Its
creation of nodes that align more precisely with the dataset disadvantage is that it is suitable for small data sets and noise
than those generated by traditional DT methods. To validate
has a relatively large influence on it. When the training data
the robustness and precision of developed SDT, and evaluated
it across four distinct datasets. However, the developed set is large, the decision tree will become very complicated.
method was utilized to integrate a new pruning method into All samples in this type of interval are 0 or the maximum.
the SDT algorithm to tackle overfitting problem. The area containing all samples in this type of range is called
a "class" field set, that is, all samples in the sample point set
have the same attributes. The core of ID3 algorithm is that
the ID3 algorithm we obtained mainly uses parameter
estimation and linearization, so it can be realized by
appropriately modifying the relationship between data and

orized licensed use limited to: National Hellenic Research Foundation & National Documentation Centre. Downloaded on September 25,2024 at 13:11:13 UTC from IEEE Xplore. Restrictions ap
 analyzing data. The parameter estimation algorithm is based mathematical analysis, we can well understand whether the
on the most classical parameter optimization technology in attributes and characteristic values of an object or a class of
the field of mathematics, that is, the most widely used and objects meet the expectations, and calculate the possible
studied parameter optimization technology. Because of the change trend and direction of the object in the future
according to the model. In the process of data analysis,
complexity of mathematics, the amount and complexity of
decision tree is very important and necessary. It can transform
theoretical knowledge, and the algorithm problem is not well a complex, multivariable, strongly coupled problem into
solved, we adopt some simple methods to improve this multiple independent but closely related unordered
shortcoming. Set s as the set of all s data samples, and relationships, without paying attention to a class of factors as
determine the rules according to the decision tree model. If a classifier. In mathematical analysis, we often encounter
we establish rules without any restrictions, we will get many some data classifications, but these classifications are
samples of different types. Therefore, in order to make the standard and purposeful. However, for an undetermined
data to be processed as little as possible close to the actual format, the relationship between different attributes and the
results or reduce the workload so as to reduce the calculation intrinsic function relationship are related. In order to realize
pressure. the model, it is necessary to identify the variables studied. If
they belong to the same category, they can be expressed as
( , ,… ) = −∑ log ( )
having some similarity. In the process of data processing,
(1)
system design and algorithm implementation, we can use
some different types of decision tree classification models to
Where pi = Si / S is the probability that any sample support the specific analysis problems we need. These
belongs to CI. Note that the logarithmic function is based on methods are applicable to ID3 decision trees. The format
2 because the information is encoded in binary. represents the internal relationship and extension
characteristics between variables. It is one of the most
E(S) = − P log (P ) (2) important tasks for us to classify (i.e., make decisions) each
type of thing when it has certain laws, characteristics or
characteristics to meet its research needs; However, it is
Where PIJ = SIJ / SJ is the probability that the sample in impossible to directly describe these information in actual
SJ belongs to CI. The information gain obtained by branching data, which makes it necessary to convert them into
on attribute a is: mathematical models that can be used.

Gain(A) = I(s , s , … , s! ) − E(A) (3) C. Problems of ID3 Decision Tree Classification Algorithm
1) Because each decision tree: table is not perfect and
ID3 algorithm has the advantages of clear theory, simple incomplete, and there may be some deficiencies, these
method and strong learning ability. problems make the model unable to be established, which
3) Decision tree classification rules: Decision tree affects the accuracy of the results. When determining the
classification is to discretize the attributes of continuous parameters, the algorithm is often not perfect because of the
variables according to their internal attributes (i.e. large amount of data.
"indicators"), and then form a class. This class is used to 2) The ID3 algorithm does not return a search: The most
describe the relationships and internal laws between things important feature of ID3 algorithm is that it will not return
in the system. Then the function model is established by search results. Whenever a property is selected for testing at
mathematical method. Decision tree classification is a class a level in the tree, it does not return to reconsider the
based learning method. Its main idea is to divide data into selection. In this way, it is easy to make the algorithm
attribute categories. Each decision tree represents a set of converge to the local optimal response rather than the global
attribute mapping relationships with other categories. All optimal response.
possibility sets in this set are called "trees". Mathematically 3) ID3 algorithm is a greedy algorithm: For incremental
speaking, we can reduce a problem to multiple sub samples learning tasks, the original decision tree should be deleted
to form a component model of the problem, and give the every time an instance is added, and a new decision tree
function values it contains and the corresponding needs to be rebuilt because it cannot receive learning
eigenvalues of each sub sample; Another example is that instances incrementally. Therefore, ID3 algorithm is not
each decision tree is taken from the data layer to represent it. suitable for incremental learning.
In the actual application process, there are often many IV. EXPERIMENT
problems that need to be solved with different types of data.
We usually choose the most appropriate algorithm to A. Mathematical Data Analysis and Construction Based on
partition data samples. ID3 Decision Tree Classification Algorithm
1) ID3 decision tree classification is based on data.
B. Function of ID3 Decision Tree Classification Algorithm
on Mathematical Data Analysis During analysis, the following steps can be used: take the
attributes of each user as input variables. Then determine the
In our life, many problems are inseparable from data, such objective function value according to the required sample
as transportation, aviation and communication. This
information is usually one of the important problems that size. This method is used to calculate the relationship
people need to study and solve, some of which are the themes between two events, which are described as high or low
generated when describing an event, or the conclusions drawn similarity relative to the actual problem and the known
when predicting the event itself. Therefore, through situation. It can also be used in the decision-making process

orized licensed use limited to: National Hellenic Research Foundation & National Documentation Centre. Downloaded on September 25,2024 at 13:11:13 UTC from IEEE Xplore. Restrictions ap
 under special circumstances. It is also used to study whether work better rather than affect its normal operation. Therefore,
changes in specific types of data objects will affect the we can divide these factors into two categories: one is to
prediction results in certain scenarios when other conditions obtain the required raw materials and finished products from
remain unchanged. In actual production and life, data mining the target factory and determine the next process time
technology can describe a large number of uncertain and according to the output; the other is to use data to predict the
random problems, and decision tree algorithm is a very extent of the next step to make decisions. This model
effective, important and convenient choice with good represents the possible results of these objects under different
application prospects. ID3 decision tree classification conditions by constructing a "feature" or "marker function",
method is used to deal with the similarity between multi and obtains the characteristics and state description
attribute sample classes. Because the basic principle and information contained in them on this basis, so that
steps of ID3 algorithm are the same, the decision analysis prediction analysis and research can be carried out according
based on the actual data can draw the desired conclusions. to their characteristics. As shown in Figure 1, the
Therefore, it is necessary to make a reasonable judgment on mathematical data analysis framework includes five
the three parts of ID when making decisions. The strict modules: data input, data catalog, data query, data download
control of product volume, size and other parameters in the and data statistics.
industrial production process is to enable the machine to

Fig. 1. Mathematical Data analysis framework

V. DISCUSSION
B. Performance Evaluation of Decision Tree Classification A. Performance Yest and Analysis of ID3 Decision Tree
Algorithm Classification Algorithm
1) Before decision tree classification: we must first make Table 1 shows the performance data of ID3 decision tree
a detailed, comprehensive and accurate analysis of the classification algorithm.
problem studied, including the number of samples in each
object. In this process, we can use various statistical methods TABLE I. ID3 DECISION TREE CLASSIFICATION ALGORITHM
PERFORMANCE
to evaluate its performance.
2) After determining the objective function and Test Data Process time Classification
times bulk (s) accuracy(%)
constraint criteria: we should select the appropriate penalty 1 10 1 100
factors that can best achieve the optimal effect. Then, we can 2 40 1.8 95
adjust these constraints to make them have the best 3 70 2.5 93
performance and adapt to the environment, and finally get a 4 100 3.4 93
5 150 4 93
good and satisfactory result.

orized licensed use limited to: National Hellenic Research Foundation & National Documentation Centre. Downloaded on September 25,2024 at 13:11:13 UTC from IEEE Xplore. Restrictions ap
 Fig. 2. Analysis of ID3 decision tree classification algorithm performance technology. ID3 algorithm (machine learning, artificial
observation) is one of the research hotspots in intelligent
In performance evaluation, the parameter values and control field. In the field of mathematics, ID3 algorithm is a
characteristic variables of ID3 algorithm are usually widely used data analysis method. This technology can be
calculated first, and then the corresponding values are used in the following aspects, such as using sample sets for
calculated according to the formula. However, each data point decision-making, estimating model parameter values, using
has different influencing factors. This classification algorithm optimal initialization methods, etc. This paper studies and
can accurately reflect the error degree between the value of a learns a series of related knowledge, such as how to obtain
variable in other samples and the theoretical value, and take more accurate original information of mathematical data in
corresponding measures to correct the error according to the limited time, and whether the decision object is suitable for
size, so as to improve the decision-making efficiency. the estimation, judgment and selection process, After
Therefore, in order to obtain highly representative and analyzing and comparing various typical models and drawing
accurate results, it is necessary to comprehensively consider conclusions, ID3 decision tree method used in mathematical
all aspects of the problem to obtain the final correct, reliable, application is selected to realize its functions and propose a
practical and reasonable scheme that meets the requirements reasonable scheme to make it meet the actual needs. At the
of the actual situation. In the previous work, this paper same time, mathematical decision tree model is analyzed and
described the principle of ID3 decision number algorithm, performance tests are conducted to verify the feasibility and
and now tested the data processing time and classification effectiveness of the algorithm, The more accurate result is the
accuracy of this algorithm. In the test process (as shown in BP neural network structure based on parameter estimation
Figure 2), it can be found that for high-resolution decision and fuzzy theory inference.
trees, if traditional methods are used for classification, the
amount of calculation and time will increase, and when ID3 REFERENCES
decision algorithm is used to improve the model, the decision [1] Basma E. El-Demerdash, Assem A. Tharwat, Ihab A. El-Khodary: A
accuracy can be effectively improved. The analysis time of Unified Mathematical Model for Stochastic Data Envelopment
Analysis. Int. J. Serv. Sci. Manag. Eng. Technol. 12(1): 127-141
mathematical data based on ID3 decision number algorithm (2021)
increases with the increase of the number of data, but the [2] Yulia Panina, Arno Germond, Brit G. David, Tomonobu M. Watanabe:
overall time increases. At the same time, the classification Pairwise efficiency: A New Mathematical Approach to QPCR Data
accuracy rate reaches more than 93%, indicating that the Analysis Increases the Precision of the Calibration Curve Assay. BMC
algorithm has good performance. ID3 algorithm can be very Bioinform. 20(1): 295:1-295:12 (2019)
convenient and fast to collect data from the database, so as to [3] Remya S, Sasikala R. Classification of rubberized coir fibres using
ensure the consistency of data access, in general, whether the deep learning-based neural fuzzy decision tree approach. Soft
Computing, 2019, 23(18):8471-8485.
data set, or other data sets, can run smoothly, and there are
fewer errors. We can know from the experiment that [4] Shoya Iwanami, Shingo Iwami: Quantitative Immunology by Data
Analysis Using Mathematical Models. Encyclopedia of
classification method is the core problem of data mining and Bioinformatics and Computational Biology (2) 2019: 984-992
machine learning, and it has been widely used in practical [5] Parameshachari, B. D., and M. V. Chaitanyakumar. "Image Security
applications in recent years. Integrated learning technology is using SCAN Based Encryption Method." In 42nd IETE Mid-term
a popular research direction in the field of machine learning. Symposium on Telecom Paradigms-Indian Scenario, pp. 115-118.
If a group of learners are integrated in a certain way, better 2011.
learning effect can be obtained than that of a single learner, [6] Scherzer, Otmar. Handbook of Mathematical Methods in Imaging ||
which significantly improves the generalization ability of the Differential Methods for Multidimensional Visual Data Analysis.
2020, 10.1007/978-3-642-27795-5(Chapter 35-5):1-56.
learning system.
[7] Top?Rceanu A, Grosseck G. Decision tree learning used for the
classification of student archetypes in online courses. Procedia
VI. CONCLUSION Computer Science, 2017, 112:51-60.
In modern industrial production and scientific research, [8] KumarJ. Pavanjp. [email protected]. Venkata
data mining plays a very important role and is a new [email protected] Department of Computer Science and

orized licensed use limited to: National Hellenic Research Foundation & National Documentation Centre. Downloaded on September 25,2024 at 13:11:13 UTC from IEEE Xplore. Restrictions ap
 Engineering, National Institute of Technology, Warangal, Telangana,
India. Department of Computer Science and Engineering, National
Institute of Technology, Warangal, Telangana, India. Algorithmic
Aspects of Some Variants of Domination in Graphs. Analele
Universitatii "Ovidius" Constanta - Seria Matematica, 2020,
28(3):153-170.
[9] Es-Sabery, F., Es-Sabery, K., Qadir, J., Sainz-De-Abajo, B., Hair, A.,
García-Zapirain, B. and De La Torre-Díez, I., 2021. A MapReduce
opinion mining for COVID-19-related tweets classification using
enhanced ID3 decision tree classifier. IEEE Access, 9, pp.58706-
58739.
[10] Janardhanan, Veena, Arun Jose, B. D. Parameshachari, C.
Muruganantham, and H. S. DivakaraMurthy. "An efficient reactive
routing security scheme based on RSA algorithm for preventing false
data injection attack in WSN." International Journal of Computer
Science and Telecommunications 4, no. 9 (2013).
[11] Liu, Y. and Yang, S., 2022. Application of decision tree-based
classification algorithm on content marketing. Journal of
Mathematics, 2022.
[12] Ju, L., Huang, L. and Tsai, S.B., 2021. Online data migration model
and ID3 algorithm in sports competition action data mining
application. Wireless Communications and Mobile Computing, 2021,
pp.1-11.
[13] Chanmee, S. and Kesorn, K., 2023. Semantic decision Trees: A new
learning system for the ID3-Based algorithm using a knowledge
base. Advanced Engineering Informatics, 58, p.102156.
[14] Schmitt J, Occhino J A, Bakkum-Gamez J, et al. 1: Implementation
and Evaluation of Benign Hysterectomy Decision Tree Algorithm.
American Journal of Obstetrics and Gynecology, 2017, 216(3):S574-
S575.
[15] Scherzer, Otmar. Handbook of Mathematical Methods in Imaging ||
Differential Methods for Multidimensional Visual Data Analysis.
2020, 10.1007/978-3-642-27795-5(Chapter 35-5):1-56.
[16] Alaa H N, Mohamed S A. On the Topological Data Analysis
Extensions and Comparisons. Journal of the Egyptian Mathematical
Society, 2017, 25( 4):406-413.
[17] Shaheen E M, Elgamel S A. Mathematical Analyses of the GPS
Receiver Interference Tolerance and Mean Time to Loss Lock.
Defence Technology, 2019, 015(003):440-449.

orized licensed use limited to: National Hellenic Research Foundation & National Documentation Centre. Downloaded on September 25,2024 at 13:11:13 UTC from IEEE Xplore. Restrictions ap