Classification and Prediction Based Data Mining Algorithm in Weka Tool

© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 908
Classification and Prediction Based Data Mining Algorithm in Weka
Tool
Renu1, Kanika2
------------------------------------------------------------------------***-----------------------------------------------------------------------
Abstract-Process of extract unseen and hidden
information from large set of data is Data Mining. Different
techniques and algorithm are used to get the meaningful
information from the large set of data. Different
classification algorithm are used just like J48, SMO, REP
tree, Naïve Bayes, Multilayer perception to extract meaning
information from large set of dataset. Predictive data
mining that use historical data, statistical modeling, data
mining technique and machine learning to make prediction
about future outcomes. Predictive analytics used in
different area to identify risks and opportunities. Weka tool
are use to predict new data using classification and
different classifier J48,SMO,REPTree,Naïve Bayes,
Multilayer Perception are classify with dataset and find
accuracy of Multilayer perception is more efficient in
accuracy.
Keywords: Data mining, Weka tool, J48 algorithm
classification, Naïve Bayes
1. Introduction
Huge amount of data is collected daily in this information
era. Analyzing huge amount of data and extract
information from that data is necessity to achieve goals.
In data mining data cleaning, incorporating earlier
knowledge on data set and interpreting perfect solution
from the pragmatic results. Data mining[1] tool weka use
to predict new data using selling house dataset. Efficiency
of different classifier is calculated using confusion matrix
and finds multilayer perception classifier has higher
accuracy.
2. Related Technique in data mining
Different data mining techniques [3] to extract insights in
data but type of data mining technique used depends on
their data and goals. To extract information from data a
wide variety of data mining technique are employed.
 Descriptive Modeling
 Clustering
 Association
 Sequential Analysis.
 Predictive Data mining Technique
 Classification
1. Decision Tree
2. Neural network.
3. Rule Induction.
 Regression.
 Prescriptive Modeling
 Pattern Mining.
 Anomaly Detection.
3. Methodology
Weka contains a collection of classifier for data analysis
with graphical user interface for easy access. Original
non-Java version of weka was a Tel/TK front-end to
modeling algorithms implemented in other programming
languages plus data preprocessing utilities in C and a
make file based system.Orignal version was design as a
tool for analyzing data from agriculture domains. Weka3
java based version developed in 1997 is used in different
application areas particularly for education purposes and
research. Several standard data mining tasks data
preprocessing, clustering, classification, regression,
visualization and feature selection supported by
weka.Input to weka is expected to be formatted according
to the attributed relational file format.
Figure 1 Weka Data Mining Tool
4. Collect Dataset and preprocessing
Collection of related items of related data accessed
individually is dataset. Process of preparing the raw data
and making it suitable for a machine learning model just
like apply filter and convert file into arff, handling missing
data etc is data preprocessing. Used data in the paper is
collected from kaggle.com.
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 06 | Jun 2023 www.irjet.net p-ISSN: 2395-0072

Figure 2 Dataset of house
5. Predict new data based on Dataset and
Classifier
In prediction [4] use Dataset housing and classifier J48 by
supplied
Training data as dataset and Supplied test data to predict
unknown attribute.
Figure 3 Predict new data j48 Classifier
6. Performance evaluation
Different machine and deep learning measurement can be
applied on the various classifier models. The
measurements are Accuracy, Recall and Precision is the
important criterion used to assess a model performance.
The value of the confusion matrix which is generated
during the testing of the model is considered to calculate
those measurements. A confusion matrix is N*N matrix
used for evaluating the performance of classification
model. After classification confusion matrix compares the
actual target values with predicted by the machine
learning model. Confusion matrices give a better idea of a
model performance.
Accuracy=Total correctly classified/Actual
Precision=Corrected predicted/Total predicted
Recall=correctly classified/Actual
6.1. Classifier J48
Figure 4 Classifier J48
Accuracy, precision, recall of Classifier J48 using
confusion matrix

Table 1 Confusion matrix J48
a b c Total
18 1 1 20
4 10 1 15
0 3 6 9
22 14 8 44
= ((18+10+6)/44)*100=77.27%
A=18/22=0.818
B=10/14=0.714
C=6/8=0.75
A=18/20=0.9
B=10/15=0.667
C=6/9=0.667
6.2. Classifier SMO
Figure 5 Classifier SMO
Accuracy, precision, recall of Classifier SMO using
confusion matrix
Table 2 Confusion Matrix SMO
a b c Total
18 2 0 20
9 6 0 15
6 2 1 9
33 10 1 44
= ((18+6+1)/44)*100 =56.81%
A=18/33=0.545
B=6/10=0.6
C=1/1=1
A=18/20=0.9
B=6/15=0.4
C=1/9=0.1
6.3. Classifier Naïve Bayes
Figure 6 Classifier Naive Bayes
Accuracy, precision, recall of Classifier Naïve Bayes using
confusion matrix

Table 3 Confusion Matrix Naive Bayes
a b c Total
15 3 2 20
4 11 0 15
5 1 3 9
24 15 5 44
= ((15+11+3)/44)*100
=65.90%
A=15/24 =0.625
B=11/15 =0.733
C=3/5 =0.6
A=15/20 =0.75
B=11/15 =0.733
C=3/9=0.33
6.4. Classifier REPTree
Figure 7 classifier REPTree
Accuracy, precision, recall of Classifier REPTree using
confusion matrix
Table 4 Confusion Matrix REPTree
a b c Total
20 0 0 20
15 0 0 15
9 0 0 9
44 0 0 44
= ((20+0+0)/44)*100=45.45%
A=20/44=0.455
B=0/0
C=0/0
A=20/20 =1
B=0/15 =0
C=0/9=0
6.5. Classifier Multilayer perception
Figure 8 Classifier Multilayer Perception

Accuracy, precision, recall of Classifier Multilayer
perception using confusion matrix
Table 5 Confusion Matrix Multilayer Perception
a b c Total
19 0 1 20
0 15 0 15
1 0 8 9
20 15 9 44
= ((19+15+8)/44)*100=95.45%
A=19/20=0.95
B=15/15 =1
C=8/9=0.88
A=19/2 =0.95
B=15/15 =1
C=8/9 =0.88
6.6. Different Classifier Analysis
Figure 9 Different Classifiers Analysis
7. Accuracy of Different Classifier
The dataset is tested and analyze with classification
algorithm [6] those are Multilayer perception, J48, Naïve
Bayes, SMO, J48 and REPTree. Comparison of accuracy of
all classifier is done it has been find that Multilayer
Perception classifier perform best with accuracy.
Accuracy is metric for evaluating classification models.
To increase the accuracy of model various method are
used. Easiest way to improve the accuracy of model is to
handle missing values. These some methods are to
increase accuracy
 Acquire more data.
 Missing value treatment.
 Outlier treatment.
 Feature Engineering.
 Applying different model.
 Cross validation.
 Ensembling methods.
 Hyperparameter tuning.
Table 6 Different classifier Accuracy
Classifier Accuracy
Multilayer Perception 95.45%
J48 77.27%
Naïve Bayes 65.90%
SMO 56.81%
Reptree 45.45%
As above Figure10 show that accuracy of Multilayer
Perception classifier is high that is 95.45% as compare to
the other classifier.
Figure 10 Accuracy of Classifier
Conclusion
In this paper classification technique J48 is used to
predict the data using housing dataset and also analysis
the various classifiers and find that multilayer perception
perform best with high accuracy.Weka data mining tool is
easy to understand and interfaced with various

technique. Hence future of data mining is promising for
further research and can be applied in different areas due
to the availability of huge databases.
References
[1] https://en.wikipedia.org/wiki/Data_mining
[2] https://medium.com
[3] Jiawei Han Michelin Kamber,”Data Mining Concepts
and Techniques”, Morgan Kaufmann Publishers
[4] M.Ramaswami and R.Bhaskaran,”A CHAID Based
performance prediction model in educational data
mining,”Journal of computer science Issues
[5] Mansi Gera Shivani goel,”Data mining techniques
methods and algorithms
[6] A Michal,”IPM developers works: IBM resource for
developers and IT”

Classification and Prediction Based Data Mining Algorithm in Weka Tool

More Related Content

Similar to Classification and Prediction Based Data Mining Algorithm in Weka Tool (20)

More from IRJET Journal (20)

Recently uploaded (20)

Classification and Prediction Based Data Mining Algorithm in Weka Tool