Comparative study of various supervisedclassification methodsforanalysing deforestation factors

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 03 Special Issue: 07 |May-2014, Available @ http://www.ijret.org 816
COMPARATIVE STUDY OF VARIOUS
SUPERVISEDCLASSIFICATION METHODSFORANALYSING
DEFORESTATION FACTORS
S.P.Rajagopalan1
, C.Lalitha2
1
Professor, G.K.M College of Engineering &Technology, Chennai, TamilNadu
2
Research Scholar, Vels University, Chennai, TamilNadu
Abstract
In this paper, various supervised classification techniques are compared and the results are demonstrated. Here the classification
techniques like Decision tree, Bayesian method, Neural Networks and Rule Based method are discussed with regards to the data
sets given. The population, built up and agriculture are the factors that play a vital role in the development of country which
directly affects economic condition. In this paper, factors such as road , population, built up development ,agriculture and
industry are considered as drivers of deforestation in the study area which is located in the Erode District of TamilNadu, India.
Keywords: Supervised Classification, Decision tree, Bayesian method, Neural Network.
--------------------------------------------------------------------***------------------------------------------------------------------
1. INTRODUCTION
There are two types of classification, supervised and
unsupervised. Supervised methods classify the data which is
known and observed by the user specifically. Unsupervised
methods are classified unknowably. The results are obtained
with the given data sets by using the WEKA. The aim of this
paper is to compare the various supervised methods by using
the factors such as demographic, built up, road and
agriculture. The primary methods used in Data mining are
Data selection, data reduction and filtration. Data mining
examines and discovers various algorithms under several
computational efficiency. It integrates machine learning,
pattern recognition, statistics, databases, and visualization
techniques into one so that the information can be extracted
from the large databases.
The tasks of data mining are association rules mining,
classification, prediction and cluster analysis. Generally
speaking, association rule mining and classification rule
mining are the most effective and efficient techniques in
data mining. Classification rule mining is used for the
prediction future objects whose class label is not known.
Recently it has been determined that primary factor for the
degradation of ecosystem is deforestation of forests.
Classification results are basis for interpretation, analysis
and modelling for various environmental and socio-
economic applications. Data mining techniques can be
applied for generating the class association rules for
analysing the deforestation. In this paper, we applied various
supervised classification techniques with our data sets.
2. WEKA
The Waikato Environment for Knowledge Analysis
(WEKA) is a tool for machine learning algorithms which
can be used for classification and clustering. In this paper,
we use decision tree methods like J48 and Random forest.
Bayes algorithm such as Naive Bayes and Neural Network
methods like Multilayer perceptron are implemented in
WEKA. We divided our data set into 10 cross validation
folds and all the methods are tested and compared according
to the data.
3. CLASSIFICATION METHODS
3.1 Decision Tree
Decision Tree (DT) worksaccording to the processing and
deciding upon attributable data. Here attributes in DT are
considered as nodes and each leaf node as a class. J48 and
Random Forest were used in our experiments. It follows a
recursivemethod for a given set of data.It searches the
attributes as Depth-first strategy. It divides the class into
several nodes and tests each node that gives the best result.It
classifies the datasets invariably .This method is not suitable
for finding anddid not show good results to the given
datasets.Accuracy value of J48 methods is compared and
shown in Table 1 and Figure1.

_______________________________________________________________________________________
Random forest is also called regression trees that induces the
data from bootstrap samples of the training data.It uses
random feature selection by induction process. Itis
comparatively gives better result than CART and C4.5.
It shows a better performance, after modelling the result.
The disadvantages of DT are focus on continues attributes,
computational efficiently with growing tree size. According
to comparison provided for different classification methods
in emotion recognition, Random Forest is the best classifier
method on that group with 5 attributes and the results are
compared and shown in Table 2 and Figure 2.
3.2 Artificial Neural Networks
3.2.1 MLP
Artificial Neural Network (ANN) is the common
classification methods in data mining. Neural Network
based classifiers, Multi Layer Perceptron (MLP) and Radial
Base Function (RBF)were used in this work. MLP is a feed
forward network that makes a model to map input data to
output data. Hidden layer in MLP can include various layers
between input and output. It classifies 3 factors gradually.
The accuracy result of MLP is shown in Table 3 and Figure
3.
3.2.2RBF
RBF is another type on ANN. The input is linear and the
output is nonlinear. Here it hides several non sequential
values because of random input. At first it shows a good
result for the year 1990 and 2010, but not for 2000.So that
this algorithm is not suitable for the given datasets. The RBF
networks are divided in two feed-forward layer is shown in
Table 4 and Figure 4.

_______________________________________________________________________________________
3.3 Bayesian Methods
Bayesian methods are one of the solution for the
classification methods in data mining. In our work, Naive
Bayesis implemented for classification. It follows an
independent feature model with strong independence
assumptions. This method is applicable for the statistical
data.
Classification is done by appropriately to the attributes
Cropand built up. It classifies and shows the better
performance to 10% for all the years. This method shows
good results for two attributes and is displayed in Table 5
and Figure 5.
3.4 Rule Based Classification
Decision trees can be translated into a set of rules by
creating a separate rule for each path from the root to a leaf
in the tree. However, rules can also be directly induced from
training data using a variety of rule-based algorithms.
Classification accuracy of rule learning algorithms can be
improved by combining features using the background
knowledge of the user automatic feature construction
algorithms.
3.4.1 1R Algorithm
One Rule is a simple accurate, classification algorithm that
generates one rule for each predictor in the data and then
selects the rule with the smallest total error as its "one rule".
It is the simplest rule-based classification learning algorithm
for discrete attributes. It shows a gradual increase in the
years 1990 and 2000 for all the attributes but suddenly no
changes for the year 2010 because OneR produces rules
only slightly less accurate .The result of this method is
shown in Table 6 and Figure 6.
3.4.2 Prism
Prism is a greedy algorithm that finds a minimum spanning
tree for a connected weighted undirected graph. This means
it finds a subset of the edges that forms a tree that includes
every vertex, where the total weight of all the edges in the
tree is minimized. It can also be used to find the minimum
spanning forest and reflects the same result for all the three
decades , is in Table 7 and Figure 7.

_______________________________________________________________________________________
4 COMPARATIVE RESULTS
The Machine learning techniques such as Naïve Bayes,
Bayes network, J48, Random Forest, Multi Layer Perceptron
(MLP) and Radial Base Function (RBF) ,one R, PRISM
were used for simulation. Here we split our original dataset
of 405 samples into 66% for training purpose and remaining
34% for testing purpose. Weka incorporates k-fold cross-
validation, in which the original sample is randomly
partitioned into k subsamples. The cross-validation process
is then repeated into several subsamples .Here ,we have used
10-fold cross validation .Kappa gives a numerical rating of -
1 to 1 scale, where 1 is perfect agreement, 0 implies
expected by chance, and negative values indicate agreement
lesser than chance, Comparatively PRISM gives best result
and shows the degradation of forest. The correctly classified
Instances of various years are shown in Table8,9,10and the
accuracy rates are displayed in Figure 8,9,10.
Fig. 8-1990
Fig.9-2000
Fig. 10-2010
5. CONCLUSIONS AND FUTURE WORK
In this paper, four classifiers including Neural Network,
Naïve Bayesian, Rule based, Decision tree were tested to
determine the deforestation from the dataset of demographic
factors. All the result were classified as 1,-1 or 0. There are
many different mining and classification algorithms, and
parameter settings in each algorithm. Experimental results in
this paper are based on the default settings. Extensive
experiments with different settings are applicable in WEKA.
J48 is very simple classifier to make a decision tree, but it
gives the invariable result in the experiment. Naïve Bayesian

_______________________________________________________________________________________
classifier also showed good result only for two
attributes(Crop and Built up), but RBF classified another
two attributes(Industry and Road) properly. Rule Based
classifiers such as OneR and PRISM also showed good
result than compared with J48 or Naïves Bayesian classifier.
From this experiment, we can find that a simple Random
Forest classifier can provide best classification result for
deforestation(except one attribute).It is planned to
incorporate other techniques like different ways of feature
selection, classification using ontology.
REFERENCES
[1] PeimanMamaniBarnaghi+,VahidAlizadehSahzabi
and Azuraliza Abu Bakar “A Comparative Study for
Various Methods of Classification”, 2012
International Conference on Information and
Computer Networks (ICICN 2012).
[2] LinlinXu, Jonathan Li, Alexander Brenning“A
comparative study of different classification
techniques for marine oil spill identification using
RADARSAT-1 imagery”,Remote Sensing of
Environment 141 (2014) 14–23.
[3] .Elaine Astrand1, Pierre Enel2, Guilhem Ibos1, Peter
Ford Dominey2, Pierre Baraduc1,Suliann Ben
Hamed1* “Comparison of Classifiers for Decoding
Sensory and Cognitive Information from Prefrontal
Neuronal Populations “,PLOS ONE |
www.plosone.org.
[4] S. B. Kotsiantis “Supervised Machine Learning: A
Review of Classification Techniques”Informatica31
(2007) 249-268 249.
[5] Archana Chaudhary1, Savita Kolhe2 & Raj
Kamal3”Machine Learning Classification
Techniques: A Comparative Study “,International
Journal on Advanced Computer Theory and
Engineering (IJACTE) ISSN ISSN (Print) : 2319 –
2526, Volume-2, Issue-4, 2013.
[6] Haowen You1 and George Rumbe2 “Comparative
Study of Classification Techniques on Breast Cancer
FNA Biopsy Data” International Journal of
ArtificialIntelligence and Interactive Multimedia,
Vol. 1, Nº 3.
[7] PeimanMamaniBarnaghi+,VahidAlizadehSahzabi
and Azuraliza Abu Bakar “A Comparative Study for
Various Methods of Classification” 2012
International Conference on Information and
Computer Networks (ICICN 2012) IPCSIT vol. 27
(2012) © (2012) IACSIT Press, Singapore
[8] Gaya Buddhinathand Damien Derry “A Simple
Enhancement to One Rule Classification”.
[9] SeongwookYoun and Dennis McLeod “A
Comparative Study for Email Classification”
[10] K.Saritha1, S. Jyothi2 & K. R. Manjula3 “Class
association rule mining for analyzing deforestation
factors” International Journal of Civil,
Structural,Environmental and Infrastructure
Engineering Research and Development
(IJCSEIERD) ISSN(P): 2249-6866; ISSN(E): 2249-
7978 Vol. 3, Issue 5, Dec 2013, 237-248
[11] K.R.Manjula ,Prof. S. Jyothi, S.Anand Kumar Varma
“Analysing the factors of deforestation using GIS
“www.geospatialworld.net/paper/application/Article
View.aspx?aid=30273

Comparative study of various supervisedclassification methodsforanalysing deforestation factors

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Comparative study of various supervisedclassification methodsforanalysing deforestation factors (20)

More from eSAT Publishing House (20)

Recently uploaded (20)

Comparative study of various supervisedclassification methodsforanalysing deforestation factors