International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-5, January 2020
2790
Retrieval Number: D8650118419/2020©BEIESP
DOI:10.35940/ijrte.D8650.018520
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Paddy Crop Production Analysis Based on SVM
and KNN Classifier
Pankaj Bhambri, Inderjit Singh Dhanoa, Vijay Kumar Sinha, Jasmine Kaur
Abstract: In earlier times, the people used to fulfill their own
requirements by cultivating the crops in their own land regions.
In the economy of a nation, an important role is played by the
farming sector. A variety of fungal and bacterial infections infect
various crops. Reducing the use of insect killers is a prominent
demand of sustainable development. The minimum use of
pesticides saves environment and increases the quality of crops.
To improve the accuracy of paddy production prediction the KNN
is implemented for the paddy production prediction in data
mining. The SVM classifieris also implemented which is
compared with the KNNclassifier. The presented and earlier
classifier will be applied in python and it is expected that
accuracy will be improved and execution time will be reduced. It
is analyzed that KNN performs well as compared to SVM
classifier for the paddy production prediction as per the obtained
analytic results.
Keywords: Paddy Crop Production, SVM Classifier, KNN
classifier.
I. INTRODUCTION
Data Mining is easy and effective to process the sorted
information. Most of the large organizations use data mining
such that the issue of including huge amount of data in
warehouses can be resolved by applying simple methods.
There are several data mining tools applied in the software
for generating and automated analysis process. Furthermore,
it is important to use historical data such that new
information can be extracted. Thus, the use of software is
reduced through the automated analysis process. The time
and cost can be reduced using this software. Since the
application fields can easily use this technology for analysis,
all the complex issues are resolved by it [1]. Thus, the major
objective of applying this technique is sorting the data
available in unorganized manner. The scientific data,
military intelligence, business transactions and satellite
pictures are collectively used for generating such large rich
source of data. From such collective raw data, extracting
only the required data is necessary. For data mining,
knowledge discovery is prominent. Theiterative process is
applied here for extracting important information.
The predictive analytics decides the future of data mining.
The terms data mining and data extraction seems like
similar. However, there is an important difference between
these two terms. In data extraction, the data is obtained from
one data source and loaded into a targeted database.
Revised Manuscript Received on January 15, 2020.
Dr. Pankaj Bhambri, Dept. of Info. Tech., Guru Nanak Dev
Engineering College, Ludhiana, Punjab, India. E-mail:
pkbhambri@gndec.ac.in ,
Dr. Inderjit Singh Dhanoa, Dept. of CSE, Guru Nanak Dev
Engineering College, Ludhiana, Punjab, India. E-mail:
cecm.cse.vk@gmail.com ,
Dr. Vijay Kumar Sinha, Dept. of CSE, Chandigarh Engineering
College, Mohali, Punjab, India. E-mail: inderjit26@gndec.ac.in ,
Er. Jasmine Kaur, Dept. of CSE, IKGPTU, Jalandhar, Punjab, India.
E-mail:jasminekaur@gmail.com
Therefore, the data can be extracted from a source or legacy
system for its storage into a standard database or data
storehouse.
However, in data mining, the unclear or hidden predictive
information is extracted from big databases or data
storehouses. Further, data mining as knowledge discovery is
used to search patterns in data warehouses [2]. Here,
computational methods from statistics and pattern
identification are utilized by data mining. Therefore, the
nature of data mining is defined by the searching of patterns
in data. Different data mining tools and techniques are used
to build a predictive analytical model. Large databases are
accessed to extract data in primary step. Here, the crop yield
needs to be forecasted and the ideal condition needs to be
identified such that high yield of paddy production can be
performed [3]
K-Nearest neighbor is a lazy learner technique. This
algorithm depends on learning by analogy. It is a supervised
classification method. This classifier is used extensively for
classification purpose. This classifier waits till the last
minute prior to build some model on a specified tuple as
compared to earlier classifiers. The training tuples are
characterized in N-dimensional space in this classifier. This
classification model looks for the k training tuples nearest to
the indefinite sample in case of an indefinite tuple. Then,
this classifier puts the sample in the closest class. This
algorithm can be implemented easily. This algorithm
performs quickly in case of small data sets. However, this
algorithm performs slowly on huge amount of data and big
size data. This approach is responsive to the value of k [4].
The performance of the classification model also gets
affected by this.
Support vector machines classification model gives good
performance on unknown data. The maximum margin
classification model is the simplest example of this
algorithm. This classification model provides solution of the
most fundamental classification issue. This issue is known
as binary classification with linear separable training data.
This classification model finds hyperplane with the maximal
margin. In support vector machine, some slack variables are
established to manage the nonlinear separable cases. Some
training errors could be handled using this phenomenon.
This reduces the effect of noise in training data. Through the
selection of uppermost probability, classification is
executed. This classifier involves a penalty metric that
permit a definite amount of misclassification. This is mainly
imperative for non separable training sets [5].
II. RELATED WORK
UraiwanInyaem, et al., (2018) stated that various
investigations had been performed in the domain of data
mining.