To Development Manufacturing and Education using Data Mining A Review

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 3 Issue 5, August 2019 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD27910 | Volume – 3 | Issue – 5 | July - August 2019 Page 2168
To Development Manufacturing and
Education using Data Mining: A Review
Aye Pwint Phyu, Khaing Khaing Wai
Department of Information Technology Support and Maintenance,
University of Computer Studies, Mandalay, Myanmar
How to cite this paper: Aye Pwint Phyu |
Khaing Khaing Wai "To Development
Manufacturing and
Education using
Data Mining: A
Review" Published
in International
Journal of Trend in
Scientific Research
and Development
(ijtsrd), ISSN: 2456-6470, Volume-3 |
Issue-5, August 2019, pp.2168-2173,
https://doi.org/10.31142/ijtsrd27910
Copyright © 2019 by author(s) and
International Journal ofTrend inScientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the Creative
Commons Attribution
License (CC BY 4.0)
(http://creativecommons.org/licenses/by
/4.0)
ABSTRACT
In modern manufacturing environments, vast amounts of data arecollectedin
database management systems and data warehouses from all involved areas.
Data mining is the nontrivial extraction of implicit, previously unknown, and
potentially useful information from data. It is the extraction of information
from huge volume of data or set through the use of various data mining
techniques. The data mining techniques like clustering, classification help in
finding the hidden and previously unknown informationfromthedatabase.In
addition, data mining also important role and educational sector. Educational
Data Mining (EDM) is a field of analysis and research where various data
mining tools and techniques are used tooptimizetheapplications ineducation
sector. The paper aims to analyze the enormous data from the education
sector and provide solutions and reports for specific aspects of education
sector such as student’s performance and placements. Moreover, this paper
reviews the literature dealing with knowledge discovery and data mining
applications in the broad domain of manufacturing with a special emphasis on
the type of functions to be performed on the data. The major data mining
functions to be performed include characterization and description,
association, classification, prediction, clustering and evolution analysis.
KEYWORDS: Data Mining, EDM, Manufacturing, Literature review
1. INTRODUCTION
In most sectors, manufacturing is extremelycompetitiveand
the financial margins that differentiate between success and
failure are very tight, with most established industries
needing to compete, produce and sell at a global level. To
master these trans-continental challenges, a company must
achieve low cost production yet still maintain highly skilled,
flexible and efficient workforceswho areabletoconsistently
design and produce high quality and low cost products. In
higher-wage economies, this can generally only be done
through very efficient exploitation of knowledge (Harding
and Popplewell 2006; Choudhary et al. 2006). In modern
manufacturing, the volume of data grows at an
unprecedented rate in digital manufacturing environments,
using barcodes, sensors, vision systems etc.
The huge amounts of data in manufacturing databases,
which contain large numbers of records, with many
attributes that need to be simultaneously explored to
discover useful information and knowledge, make manual
analysis impractical. All these factors indicate the need for
intelligent and automated data analysis methodologies,
which might discover useful knowledge from data.
Knowledge discovery in databases (KDD) and data mining
(DM) have therefore become extremely important tools in
realizing the objective of intelligent and automated data
analysis. Data mining is a particular step in the process of
KDD, involving the application of specific algorithms for
extracting patterns (models) from data.
1.1 Data Mining for Manufacturing
Knowledge discovery in databases (KDD) and data mining
(DM) have therefore become extremely important tools in
realizing the objective of intelligent and automated data
analysis. The additional steps in the KDD process, such as
data preparation,data cleaning,data selection,incorporation
of appropriate prior knowledge andproperinterpretationof
the results of mining, ensure that useful knowledge is
derived from the data (Mitra etal.2002).thesefields provide
specific data mining tools that can be used invarioussteps of
a KDD process. Recently, with the growth of data mining
technology, researchers and practitioners invarious aspects
of manufacturing and logistics have started applying this
technology to search for hidden relationships or patterns
which might be used to equip their systems with new
knowledge. Early applications of data mining were mostly
applied to financial applications, for example Zhang and
Zhou (2004) described data mining in the context of
financial applications from both technical and application
perspectives. In this area, the competitive advantage gained
through data mining included increased revenue, reduced
cost, much improved market place responsiveness and
awareness. A recent sur-vey carried out by Harding et al.
(2006) and a special issue published on “data mining and
applications in engineering design, manufacturing and
logistics” (Feng and Kusiak 2006) clearly indicated the
potential scope of data mining in these areas to achieve
competitive advantages. A major advantage of data mining
over other experimental techniques is thattherequireddata
IJTSRD27910

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
for analysis can be collected during the nor-mal operationof
the manufacturing process being studied. Therefore, it is
generally not necessary to specially dedicate machines or
processes for data collection.
1.2 Data Mining for Manufacturing Literature Review
Han and Kamber (2001) classified data mining systems
based on various criteria such as kind of database mined,
kind of knowledge mined, kind of technique utilized and
application areas adopted. Pham and Afify (2005) reviewed
machine learning techniques in the manufacturing domain..
Harding et al. (2006) surveyed data mining systems in
different application areas of manufacturing,includingsome
less considered areas such as manufacturing planning and
shop floor control. However, in the last few years, data
mining research in manufacturing has increased at an
exponential rate. Han and Kamber (2001) mentioned that
the kind of knowledge to be mined determines the data
mining functions to be performed. Possible kinds of
knowledge include concept description(characterization and
discrimination), association classification, clustering, and
prediction. The aim of this paper is therefore to consolidate
the existing state-of-the art research efforts concerning the
current practices in data mining applications in
manufacturing based on the kind ofknowledgeminedand the
kind of technique utilized, thereby identifying promising
areas for study. The remainder of the paper is organized as
follows. briefly discusses about KDD, data mining, and the
kinds of knowledge that particularly occur inmanufacturing
contexts. Section “Concept description(characterization and
discrimination) in manufacturing” will discuss concept
descriptions which include characterization and
discrimination in manufacturing. Classification in
manufacturing is discussed in section “Classification in
manufacturing,” followed by clustering in manufacturing in
section “Clustering in manufacturing”. Section“Predictionin
manufacturing” discusses prediction in manufacturing, and
association in manufacturing is discussed in section
“Association in manufacturing”. Details of our novel text
mining approach are given in section “Detailed analysis and
discussion: a text miningperspectiveonreviewedliterature”
and this is followed by conclusions in section “Conclusion”.
1.2.1 KDD, data mining and knowledge types
KDD is the nontrivial process of identifying valid, novel,
potentially useful, and ultimately understandable patterns in
data (Fayyad et al. 1996a). The KDD process is interactive
and iterative involving more or less the following steps
(Fayyad et al. 1996b; Mitra et al. 2002).
 -Understanding the manufacturing domain
 -Collecting the targeted data
 -Data cleaning, pre-processing and transformation
 Data integration
 -Choosing the functions of data mining
 -Choosing the appropriate data mining algorithm
 -Data mining
 -Interpretation and visualization
 -Implementation of discovered knowledge
 -Knowledge storage, reuse and integration into the
manufacturing system
Data mining is an interdisciplinaryfieldwith thegeneral goal
of predicting outcomes and uncovering relationships indata.
It makes use of automated tools and techniques, employing
sophisticated algorithms to discover hidden patterns,
associations, anomalies and/or structure from large
amounts of data stored in a data warehouse or other
information repositories. In the context of manufacturing,
two high level primary goals of data mining are prediction
and description. Descriptive data mining focuses on
discovering interesting patterns to describe the data.
Predictive data mining focuses on predicting the behaviour
of a model and determining future values of key variables
based on existing information from available databases.The
boundaries between, descriptive and predictivedatamining
are not sharp, e.g. some aspects of the predictive model can
be descriptive, to the degree that they are understandable
and vice versa. The goals of prediction and description can
be achieved by using a variety of data mining tools and
techniques. The next section there-fore describes a range of
functions and reviews their applicability in manufacturing
domains.
1.2.2 Concept Description (Characterization and
Discrimination) in Manufacturing
Characterization can be used to identify the features that
significantly impact the quality. Characterization provides a
concise and succinct summarizationofthegivencollectionof
data, while concept or class discrimination or comparison
provides descriptions that compare two or more collections
of data. In manufacturing contexts, these functions are
basically used to understand the process. Huyet (2006)
proposed an evolutionary optimization and data mining
based approach to produce the knowledge of systems
behaviour in a simulated job shopbased productionprocess.
Assigning proper dispatching rules is an important issue in
enhancing the performance measures for a flexible
manufacturing system(FMS).LeeandNg(2006)presented a
hybrid case based reasoning (HyCase) system for online
technical support of PC fault diagnosis. Romanowski and
Nagi (1999) applied a decision tree based data mining
approach on a scheduled maintenance dataset and a
vibration signal dataset. Subsystems which are most
responsible for low equipment availability arerecognized in
the scheduled maintenance data and a recommendation for
preventive maintenance interval is made.
1.2.3 Classification in Manufacturing
Classification is a useful functionality in many areas of
manufacturing, for example, in the semiconductor industry,
defects are classified to find patterns and derivetherules for
yield improvement. Online control chart patternrecognition
(CCPR) is another example of classification for SPC, because
unnatural patterns displayed by a control chart can be
associated with specific causes that adversely impact the
manufacturing process. Classification is a learning function
that maps (classifies) a data item into one of several
predefined categorical classes. Generally, classification is
performed in two steps. In the first step, a model is built to
describe a predeterminedsetofdata classes orconcepts,and
this is done by analyzing the database tuples described by
attributes, which collectively form the training dataset.
Rokach and Maimon (2006) applied a feature set
decomposition methodology for quality improvement. They
developed the Breadth Oblivious Wrapper(BOW)algorithm
and showed its superiority over existing tools on datasets
from IC fabrication and food processing. The idea is to find
the classifier that is capable of predicting the quality
measure of product or batch based on its manufacturing
parameters. Braha and Shmilovici (2002) presented three
classification based data mining methods (decision tree
induction, neural network and composite classifier) for a

new laser based wafer cleaning process called advanced
wafer cleaning. The purpose of the data mining based
classifier is to enhance understanding of the cleaning
process by categorizing the given data into a given
predefined number of categorical classes and determine to
which the new data belongs. A fractal dimension based
classifier was proposed by Purintrapiban and
Kachitvichyanukul (2003) for detection of unnatural
patterns in process data. Kusiak (2002a), Kusiak (2002b)
applied data miningtosupportdecisionmakingprocesses by
using different data-mining algorithms to generate rules for
a manufacturing system. A subset of these rules was then
selected to produce a control signature for the
manufacturing process where the control signature is a set
of feature values or ranges that lead towards an expected
output.
From this review, the major application areas where data
mining tools and techniques are used for classification
include fault diagnosis, quality control and condition
monitoring. In order to perform the classification task,
decision tree, rough set theory, hybrid neural network and
other hybrid approaches have been successfully used. In
hybrid approaches, Fuzzy logic is used often in combination
with other techniques to deal with noise and uncertainty in
the data. The next section will deal with clustering and its
performance on manufacturing databases.
1.2.4 Clustering in Manufacturing
Clustering is an important data mining function that can be
performed on specified manufacturing data such as order
picking in logistics and supply chain. For example order
picking is routine in distributioncentersandbeforepickinga
large set of orders, orders are clustered into batches to
accelerate the product movement within the storage zone.
Clustering is also useful in the formation of cells in cellular
manufacturing where it is used for the simultaneous design
of the part families and machine cells.
Clustering is also known as unsupervised learning. Unlike
classification (supervised learning), in clustering the class
object of each data object is not known. Clustering maps a
data item into one of several clusters, where clusters are
natural groupings of data items based on similarity metrics
or probability density models (Mitra et al. 2002;
XuandWunsch2005). Sebzalli andWang (2001) applied
principal component analysis and fuzzy c means clustering
to a refinery catalytic and fuzzy c means clustering to a
refinery catalytic and fuzzy c means clustering to a refinery
catalytic and fuzzy c means clustering to a refinery catalytic
process to identify operational spaces and develop
operational strategies for the manufacture of desired
products and to minimize the loss of product during system
changeover. Kim and Ding (2005) proposed a data mining
aided optimal design method for fixture layout in a four
station SUV side panel assembly process. Clustering and
classifications are carried out to generate a design library
and design selection rules, respectively. Torkul et al. (2006)
showed the outperformanceoffuzzycmeans clusteringover
crisp methods on a selected data set. Romanowski and Nagi
(2001) proposed a design system which supports the
feedback of data mined knowledge from life cycledata to the
initial stages of the design process. Romanowski and Nagi
(2005) and Romanowski and Nagi (2004) also applied a
data-mining approach for forming generic bills of
materials(GBOMS) entities that represent the different
variants in a product family and facilitate the search for
similar designs and the configurations of new variants. Lee
et al. (2001) proposed an intelligent inline measurement
sampling method for process excursion monitoring and
control in semiconductor manufacturing. The average
diagnostic accuracy of 80% showed thatthishybrid model is
promising for an EMI diagnostic support system.Huiand Jha
(2000) investigated the application of data mining
techniques to extract knowledge from the customer service
database for decision support and fault diagnosis.
Predictability of manufacturing processes, quality,
maintenance, defects, or evenwithinmanufacturingsystems
is of vital importance. For example in the context of
maintenance, predictions can be made aboutwhatcondition
maintenance will be required or how equipment will
deteriorate based on the analysis of past data. Feng and
Kusiak(2006), Feng et al. (2006) showed that there is no
significant statistical advantage of using fivefold CV over
threefold CV and or of using a two hidden layer neural
network over a one hidden layer neural network for turning
surface roughness data. Pasek (2006) used the rough set
theory based classifier for the prediction of cutting tool
wear. For tool condition monitoring Sunetal.(2005)applied
a neural network for recognition of tool condition in a
monitoring system. Sylvain et al. (1999) used different data
mining techniques including decision trees, rough sets,
regression andneuralnetworks topredictcomponentfailure
based on the data collected from the sensors of an aircraft.
Their results also led to the design of preventive
maintenance policies before the failure of any component.
Lin and Tseng (2005) introduced a cerebellar model
articulation controller (CMAC) neural network based
machine performance estimation model. Tsai et al. (2006)
presented a case based reasoning (CBR) system using
intelligent indexingandreasoningapproaches forPCBdefect
prediction. Knowledge elicitation is a technique that is
generally used for producing rules based on human
expertise. A method was developed by Browneet al. (2006)
to fuse knowledge elicitation and data mining using an
expert system.
1.2.5 Association in Manufacturing
Association rules mining was first introducedin1993,and is
used to identify relationships between a set of items in a
database (Agrawal et al. 1993). These relationships are not
based on inherent properties of thedata themselves (as with
functional dependencies), but rather are based on co-
occurrence of the data items. In design contexts, the
associations between requirements may provide additional
information useful for the design. For example, technical
specifications might state that a car that has two doorsanda
diesel engine requires a specific speed transmission. Insuch
cases, knowing the number of cars with two doors and the
number of cars with a diesel engine is not relevantwhilst the
number of cars with two doors and a diesel engine is useful,
for example to determine the capacity of a manufacturing
process. The nature of this association can be extracted by
applying data mining algorithms on the database.
Agard and Kusiak (2004b) applied data mining to customer
response data for its utilization in the design of product
families. Jiao and Zhang (2005) developed explicit decision
support to improve the product portfolioidentificationissue
by using association rule mining from pastsales andproduct

records. This review shows that the major areas where
association as a data mining function has been applied
include product design,process control,mass customization,
cellular design etc. Association rule mining has been applied
as a dominating tool to identify the associations among
variable.
2. Data Mining For EDM
Similarly; Education sector is one of the sectors where data
mining is relatively new as compared to other sectors and
hence it is under-utilized.TheInternationalEducationalData
Mining Society defines EDM as follows: “EDM is anemerging
discipline, concerned withdeveloping methodsforexploring
the unique types of data thatcomefromeducational settings,
and using those methods to better understand students,and
the settings which they learn in” (Baker, 2015). “The EDM
process converts raw data coming from educational systems
into useful information that could potentially have a greater
impact on educational research and practice” (Romero and
Ventura, 2010). EDM by and large comprises (Baker, 2010;
and Romero and Ventura, 2010) four phases:
1. Data Collection:
The first phase of EDM is to explore the interrelations
between the data of educational sector using data mining
techniques, i.e., classification, clustering, regressionetc.This
phase focuses on grouping the data and also preprocesses
them for mining. Data size is enormous and henceneeds a lot
of preprocessing in order to obtain a desired outcome.
2. Validating Relations:
The second phase of EDM is validation of found inter-
relations between data with the goal that uncertainty can be
evaded. The relations are then validated based on the
training dataset.
3. Predicting the Future Progress:
The third phase is to make predictions forfutureonthebasis
of validated relationships in learning environment.
4. Decision Making:
The fourth phase is utilizing the gathered information and
making calculated decisions usingtechniques likeprediction
and classification.
Educational institutes use data mining techniques for
purposes like analyzing and visualization of data, predicting
student’s performances. Data mining techniques like
clustering can be used to group students based on the
parameters decided by the analyst. Data mining helps in
identifying unwanted behaviors provides feedback to
instructors on student’s performance with information to
support the evaluation.
Data mining utilizes many techniques and algorithms, and
they can be classified into the following categories:
 Prediction:
It aims at generating a single target attribute of the data by
analyzing all the other attributes and generating patterns
from them (Romero and Ventura, 2013). Types ofprediction
techniques are classification, clustering, etc.
 Classification:
Groups information/data into a few predefined attributes.
The techniques utilized for classification are:
 Decision tree
 Naive-biased classification
 Generalized Linear Models (GLM)
 Support vector machine etc.
 Clustering:
In clustering technique, the dataset is divided into various
groups, known as clusters. As per clustering phenomenon,
the data point of one cluster should be more similar to other
data points of same cluster and more dissimilar to data
points of another cluster. There are two ways of initiation of
clustering algorithm: Firstly, clustering algorithm has to be
started with no prior assumption; and secondly clustering
algorithm has to be started with a prior postulate.
 Relationship Mining:
It helps in finding relations between values in a data corpus
and organizing them as rules. There are various relationship
mining procedures such as association rule mining,
sequential pattern mining, correlation and causal data
mining. In EDM, relationship mining is utilized to recognize
connections between the understudy’s web exercises and
the last outcomes and to display student’s critical thinking
movement successions.
 Discovery with Models:
It uses an approved model of a method utilizing expectation,
grouping, or information buildingas a segment aheadoftime
examination, for example, forecast or relationship mining.It
is utilized as a part of circumstances to get a kick out of the
chance to recognize the connections between the
understudy’s history and qualities.
 Outlier Detection:
The point of outlier detection is todistinguishcharacteristics
that are unfathomably interesting than whatever is left of
information. An exception is an alternate occasion that is
normally more prominentor lesserthanalternate esteems in
information corpus. In EDM, exception identification can be
used to recognize varieties in the students or instructor’s
activities or practices, unpredictable learning forms, andfor
Distinguishing understudies with learning troubles
(Dominguez et al., 2010; and Baker, 2015).
2.1 Data Mining for EDM Literature Review
Pruthi and Bhatia (2015) utilized the data mining technique
to predict the student’s performance in the placement
activity of the computer science students and also predict
the company they are going to be placed in (name and type
of company). They used the classification process based on
the parameters like their overall resultandspecific student’s
marks. The main issue with their process was that they used
a limited amount of data only available with the University
for the training and testing purposes. They identified the
parameters as marks in many cores IT subjects.
Dominguez et al. (2010) developed a process and feedback
generation engine that generated feedback based on the
current performance or the performance of similar class of
students. They used student information, current
performance and the performance history of other users as
parameters to predict the performance of the student and
thus provide real-time feedback for them. This is a real-time
generation of the evaluation of student andhenceprediction
of student performance. Educational data mining methods
are based on statistics, machine learning and database
theory. The main activities of this area are: data mining
usage for Intelligence Tutoring Systems support, analysis of

education processes, visualdata miningandvisual education
process pattern. The analysis ofthescientificliteratureinthe
field of using the methods of data mining showed that this
problem is interesting to many modern researchers. For
example, in (Ceylan 2015) the authors propose a searching
model system related to student success in the form of
classifiers, each of them is learned with different dataset
with hundreds of thousands of lines in relation to sections.
Received classifiers would serve as an advisory system for
students who want to choose courses prior the registration
in the semester. In ( Herlina 2017), the role of the K-Means
algorithm for classifying students learning activities using e-
learning was showed. This algorithm helped toformstudent
activity and improving student abilities cluster.Anapproach
based on minimal spanning tree for clustering e-learning
resources is proposed in (Wu 2016). The developed
clustering method can classify students into groups sothata
homogeneous classification can increase the learning
effectiveness.(Rawat 2019) justified the use of cluster
analysis for classifying a new studentintothecorresponding
class and recommending relevant courses using various
evaluation metrics. In addition, global trends, dynamic
environment, difficulty of the problems requiring greater
efficiency, adaptability, integration and coordinationofall of
relevant design process and implementation of the e-
learning systems.
2.2 Factors Affecting EDM
There are many factors affecting the aspects of EDM. The
main issues that EDM focuses on are placement, admissions,
and branch or career selection and student performances.
There are many factors that affect the areas of education.
Although there are many factors, almost all of them can be
classified into the following factors:
 Interest of Student:
Career of an individual depends on the choices he makes.
These choices are above averagely influenced by theinterest
of the student towards any area. The area in which a student
has interest in can help him perform better in terms of
academics as well as in his corporate life. If a student
chooses to pursue an occupation or academics in a topic
which he is not interested in, it can lead to a difficult life as
he would not be performing well. Interest also includes his
habits and hobbies. For example, if a person has a hobby of
traveling around, then he can choose his future in that field.
Hence interest, hobbies and habits can affect above
averagely all the factors of education.
 College Facilities:
College facilities are the thingsthatstudent pays feefor—for
better infrastructure, better faculties, library, residential
facilities, food availability and other things. All this combine
to make the basic need of a student for education. College is
responsible for providing all these facilities along with
academic knowledge which is their primary work.
 Schooling:
Children with good schooling present good academic results
in higher education. They have experienced an educational
environment that takes more interest inthepractical viewof
studies. They tend to be more mature and regular in their
assignments. Most of the students adapt to the learning
material and methodology quickly with ease. Similarly,
schooling, medium and tutoring are imperative as students
with English medium foundation generally make more
inquiries amid showing learning process. These students
who are actively participatingintheclassrooms activity tend
to have a strong base of technical and nontechnical skillset
which helps them in placement activities. Such parameters
as medium of schooling and student’s skillset help in
predicting the performance of students in the academics.
 College Reputation:
Google’s CEO is from IIT and even Microsoft’s CEO is from
Manipal University. Why not from a normal institute? Yes,
that is due to the reputation which these institutions have
made. Hence, even if a student is an above average student
from a local, non-reputed college, no MNC is going to offer a
job directly via college campus. That is just because the
college is not reputed.
3. Discussion
The reviewed literature shows that thereis a rapidgrowth in
the application of data mining in manufacturing,particularly
in the semiconductor industry. In this research, we have
briefly discussed data mining concept and its techniques for
development of knowledge management in organizations.
The next section discusses the text mining experiments
undertaken using the abstract and keywords of the 150
published works reviewed in this paper.
 Knowledge discovery in text and text mining
applications on the literature review. Following the
definition of KDD by Fayyad et al. (1996a), Karanikas
and Theodoulidis (2002) defined KDT as “the non trial
process of identifyingvalid,novel,potentiallyuseful,and
ultimately understandable patterns in unstructured
data”. Text Mining (TM) is also a step in the KDTprocess
consisting of particular data mining and natural
language processing algorithms that under certain
computational efficiency and limitations produce a
particular enumeration of patterns over a set of
unstructured textual data. KDT in reviewed literature
mainly consists of three steps as follows:
A. Abstract and keyword collection:
In our experiments, the abstracts and key words of the
literature reviewed in this paper have beencollected. Where
necessary, additional key words have also been identified
from the papers and added to the abstract for text mining.
This is important as the published abstracts often did not
include full details of the type of data mining function(s)and
areas of application discussed in the paper.
B. Retrieving and pre-processing documents:
Abstracts have only been taken from papers which deploy
data mining methodology to solve problems of
manufacturing. The additional key words have been
identified based on knowledge area,functionperformed and
technique used. The major knowledge areas examined
include manufacturing system, quality control, fault
diagnosis, maintenance, job shop, yield improvement,
manufacturing process, fault diagnosis, product design,
production control,andsupplychainmanagement.Similarly,
the functions considered include concept description,
classification, clustering, prediction and association. Major
techniques used include rough set theory, decision tree,
statistics, neural network, association rule, fuzzy c means
clustering, and regression analysis andhybridalgorithms. In
this context, the term “hybrid algorithm” indicates that
either a group of algorithms have been used in combination
to solve a particular problem, or a group of algorithms have
been used at different stages of data mining.

C. Text mining:
For the current purpose, text analysis and link analysis were
used to extract patterns, trends, useful knowledge and meet
the listed benefits. The text mining was performed as an
automatic process withmanualinterventionsduringthe pre-
processing stage. Poly analyst, which is one of the leading
data/text mining software package in the market was used
for this purpose. All the results shown and interpretations
made were automatically generated using this software. The
following subsections describe how the abovementioned
objectives were achieved. Equally, job description mining
can reveal actionable insight forstudents, employersandthe
institution. The institution can provide students with a
better understanding of co-op opportunities in various
disciplines and therefore helpthemselectthe rightacademic
program and career. Additionally, the institution may use
frequently appearing words and the clustering of jobs in
various disciplines to produce more effective promotional
material for its co-op programs and to help attract strong
students. Furthermore, students can find out what types of
jobs are available to them and what soft and technical skills
are required. In particular, clusteringcanbeusedtosegment
the job descriptions to make it easier for students to find
jobs they are interested in and institutions can align their
curricula with job market needs.
4. CONCLUSION
Knowledge discovery and data mining have created new
intelligent tools for extracting useful information and
knowledge automatically from manufacturing databases.
The present article provides a survey of the available
literature on data mining applicationsinmanufacturing with
a special emphasis on the kind of knowledge mined. The
types of knowledge identified indicatethemajordata mining
functions to be performed include characterization and
description, association, classification,prediction,clustering
in data. This paper reviewed A novel text mining approach
has been applied on the reviewed literature to identify the
popular and successful research tools and existing research
gaps, examine the under looked and overlooked areas,
identify good practices in data mining in manufacturing and
some key features unknown to data mining practitioners.
EDM and manufacturing for using data mining as an area of
research. The paper discussed various techniques, factors
and applications of EDM and manufacturing.Therearemany
factors that affect the aspects of EDM. The paper highlighted
some of them and also compared many of them based on
their impact onplacement outcomes,academicperformance,
and college and branch selection.
REFERENCES
[1] Agard, B., & Kusiak, A. (2004a). Data mining for
subassembly selection. Journal of Manufacturing
Science and Engineering, 126, 627– 631.
doi:10.1115/1.1763182.
[2] A. K. Choudhary. J. A. Harding.M.K.Tiwari.Data mining
in manufacturing: a review based on the kind of
knowledge, J Intell Manu(2009)20:501-521 DOI
10.1007/s10845-008-0145-x.
[3] Agard, B., & Kusiak, A. (2004b). Data-mining based
methodology for the design of product families.
International Journal of Production Research, 42, 15,
2955–2969.Agrawal, R., Imielinski, T.,&Swami, A. N.
(1993).Mining association
[4] Rules between sets of items in large databases. In
Proceedings of the1993 ACM SIGMOD International
Conference on Management of Data (pp. 207–216).
Washington, D.C., May 1993.Backus, P., Janakiram, M.,
Mowzoon, S., Runger, G. C., & Bhargava, A. (2006).
Factory cycle time prediction with a data-mining
approach. IEEE Transactions on Semiconductor
Manufacturing, 19(2). doi:10.1109/TSM.2006.873400.
[5] Batanov, D., Nagarur, N., & Nitikhumkasem, P. (1993).
Expert—MM: A knowledge based system for
maintenance management. Artificial Intelligence in
Engineering, 8, 283–291. doi:10.1016/0954-
1810(93)90012-5.
[6] Belz, R., & Mertens, P. (1996). Combining knowledge
based systems and simulation to solve rescheduling
problems. Decision Support Systems, 17, 141–157.
doi:10.1016/0167-9236(95)00029-1.
[7] Baker R (2010), “Data Mining for Education”, in B
McGaw, P Peterson and
[8] E Baker (Eds.), International Encyclopedia of Education,
3rd Edition, Vol. 7, pp. 112-118, Elsevier, Oxford, UK.
Baker R (2015), Educational Data Mining and Learning
Analytics, Springer International Publishing.
[9] Blagojević M and Micić Ž (2013), “A Web-Based
Intelligent Report e-Learning System Using Data
Mining Techniques”, Computers & Electrical
Engineering, Vol. 39, No. 2, pp. 465-474, Elsevier BV.
[10] Gafarova, L. M., Zavyalova, I. G., Mustafin, N. N. (2015)
‘On the features of the application of the Pearson
consensus criterion X2’, ESGI, no. 4(8), pp. 63-67.
[11] Gorlushkina, N. N., Kotsyuba, I. Yu., Khlopotov, M. V.
(2015) ‘Tasks and methods of intellectual analysis of
educational data for decision support], GTR, no. 1, pp.
472-482.
[12] Hanna, M. (2004) ‘Data Mining in the e-learning
domain’, Campus-Wide Information Systems, no. 21(1),
pp. 29-34.
[13] He, W. (2013) ‘A survey of security risks of mobile
social media through blog mining and an extensive
literature search’, Information Management and
Computer Security, no. 21(5), pp. 381-400.
[14] Herlina, Latipa Sari, Dewi,SurantiMrs.andLeni,Natalia
Zulita (2017) “Implementation of k-means clustering
method for electronic learning model” // International
Conference on Information and Communication
Technology (IconICT) IOP Publishing IOP Conf. Series:
Journal of Physics: Conference Series, Volume 930.
[15] Hussain, M. et al. (2018) “Student Engagement
Predictions in an e-Learning System and Their Impact
on Student Course Assessment Scores” //
Computational Intelligence and Neuroscience,vol 2018.
[16] Ilyina, T. S., Zakharov, N. Yu. (2016) ‘Management of
educational risks’, Vestnik VGUIT, no. 4(70), pp. 290-
295.
[17] Kamisli, Ozturk, Z., Erzurum, Cicek, Z. I. and Ergul, Z
(2017), “Sentiment Analysis an Application toAnadolu
University,” Acta Physica Polonica A, vol. 132, no. 3, pp.
753–755.
[18] Smeet M Thakrar*, Navjyotsinh Jadeja** and Nikunj
Vadher***,Educational Data Mining:A Review, The IUP
Journal of Information Technology, Vol. XIV, No. 1,
2018

To Development Manufacturing and Education using Data Mining A Review

More Related Content

What's hot (20)

Similar to To Development Manufacturing and Education using Data Mining A Review (20)

More from ijtsrd (20)

Recently uploaded (20)

To Development Manufacturing and Education using Data Mining A Review