International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169

Volume: 6 Issue: 8 05 - 07
______________________________________________________________________________________
Prediction of Crop Yield Using LS-SVM

Dr. Kamaljit Kaur Rupinder Kaur

Assistant Professor, Department of Information Technology PG Student
Guru Nanak Dev Engineering College, Ludhiana Guru Nanak Dev Engineering College, Ludhiana
Punjab, India, Punjab, India,
e-mail: [email protected] e-mail: [email protected]

Abstract— Agriculture sector is the backbone of Indian economy. Yield prediction of crop is very important problem in agriculture field.
Predicting the yield is very popular among farmer nowadays, which especially contributes to the correct choice of crop for growing. Earlier, crop
production was implemented by considering the farmer's experience on an appropriate field as well as crop. Information technology has an
important area to predict the crop production because it includes the latest advancement. In this paper, our aim is to develop the system to
achieve high accuracy and resolve the problem of yielding in agriculture. We use the LS-SVM and deep neural network learning technique to
predict the yield of crop based physical and chemical feature of plant and soil. This system helps the farmer to take the right decision about
sowing particular crop in available land.

Keywords-crop yield, LS-SVM, Deep Neural Network learning

__________________________________________________*****_________________________________________________

I. INTRODUCTION has turned out to be increasingly a part of our everyday life,

particularly for agriculture.
In India, Agriculture is the most important history. It is the
backbone of Indian economy and gives the 10-15% gross
product development to it. Yield prediction is the major Knowledge
problem in agriculture. Most of the farmers are not getting the
expectable yield of crop for various reasons. Every farmer is
excited with knowing, how much yield he is going to
anticipate. Earlier, predict the yielding of crop was
implemented through the farmer’s experience on particular Pattern evaluation
crop and field. Examine that information are accessible for
frequently to the back, where the comparing yield expectation
have been recorded. Data mining technique that training the
information is to be gathered from some time back to the past
Data mining
and accumulated information is utilized as a part of terms of
preparing which must be exploited to figure out how to
arrange future yield forecasts. Various techniques were
suggested for mining the information or data throughout the
years. The real problem in agriculture field is the information
obtaining and efficient learning exploitation. Agriculture is a Data selection and
rare crop production is dependent on various economy and transformation
weather factor. The agriculture yield of the crop is mainly
based on climate situation and soil properties. Salinity is the
most significant climate issues. Aggregation of extensive
measure of salt can decrease the productivity of yielding of
crop, effectiveness of water system, demolish soil system and
other soil properties. The dominant salts which aggregate in Data cleaning
soils are magnesium, ph range, potassium, carbonate,
phosphorus and so on. For example:- assume farming on
cotton, we could focus the farmer having less water and
human assets. Subsequently the farmer having constant supply
of water and mean temperature is applicable for paddy and Data integration
also rainfall situation additionally impact on rice cultivation
and. The amount of data is available in agriculture field and it
is used for various purposes. In this paper, the farmer’s needs a
timely instruction to predict the future yield of crop and an
Figure:- Knowledge Discovery in Database Process
analysis is to be done as to support the farmer’s to increase the
yield of crop in the agriculture field. In recent year, various
development applications have offer new requirement and In agriculture, data mining is moderately research field.
technique for information technology. Information technology Data mining is useful for analyse the large database and
provide valuable classification and pattern in the data set. The
overall purpose of the data mining technique is used to extract
5
IJRITCC | August 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 8 05 - 07
______________________________________________________________________________________
the data from a large database and transfer it into reasonable B. Deep Neural Network Learning
framework for further utilize. Deep Learning (also called organization learning) is
In this paper our main focus is to develop system that helps part of large extensive class of machine learning strategies
the farmer to take the right decision for sowing the particular depends on learning data or information representation rather
crop on available land. This work use two techniques such as than particular task algorithm. Learning can be either
LS-SVM and deep neural network learning to predict the yield supervised or unsupervised. Deep learning structure for
of crop based on available data set and resolve the problem of example deep belief networks, recurrent neural network and
yielding in agriculture. deep neural network have been employ to field consists of
II. METHODOLOGY computer version, social network separating where they
delivered result equivalent to and in some case again better
In this paper, develop the system using MATLAB than human specialists. Deep neural network learning
Software. It is the programming language implement by technique use to automatic feature extract without human
MathWorks and is simple and easy way to make learning fast interference. In this method, these systems learn to perceive
and efficient. Data is collected from various resources. Data relationship between significant features and produce excellent
mining classification techniques is used to predict the crop results.
yield in agriculture field. Second step is implemented to 1. Data collection:- Soil data and rainfall of crop is
extract the feature from database. Next step, apply the collected from Punjab Agriculture university of
classifier used to train and test the input data of crop and soil. Ludhiana and another feature of crop is collected
The model then classifies all the crop parameter and soil from online sites of agriculture. Input data is
characteristics from input value. Then it finds the type of crop, compared to original data which classifier has been
time duration and name of pesticide based on physical and conducted. Data includes the physical and chemical
chemical feature of soil and crops and show the result. feature of plant and soil.
2. Pre-Processing Phase:- Pre-processing is very
important step in data mining. In pre-processing
Pre-processing and feature phase includes insert the missing value, proper range
extraction of data and then extract the feature. The quality of
data set is very important process for analysis.
3. Feature extract:-Feature extraction requires the
simplifying the number of data that contains to
represent a huge data set. The feature of soil and crop
Pre-processing and feature are extracted in pre-processing phase is the final
extraction training set. These features include the physical and
chemical characteristics of crop and soil.
4. Processing Phase:-Processing Phase, The final data
set are formed in pre-processing are processed by LS-
Training SVM and deep neural network learning techniques.
This phase includes the training and testing input
data.
(A) Training:- Classifier is used to train the
available input data. In this phase, train the support
Testing data and preconception terms of a classifier for
function approximation and classification.
(B) Testing:- In this phase, test the trained data
and then to give the corresponding name of crop,
pesticide name and time taken.
Result
III. RESULT
The system is work on the different datasets. These
datasets include physical and chemical feature of plant and soil
Figure:-Flow diagram of different steps for yield problem such as temperature, rainfall, humidity, ph range, 5 type of soil
such as black soil, alluvial soil, red soil, laterite soil and desert
A. Least Square Support Vector Machine (LS-SVM) soil, soil moisture, iron, aluminium, magnesium, nitrogen,
LS-SVM are least square form of SVM (Support Vector potash and phosphorus. The datasets used in this system
Machine), which are an arrangement of supervised learning consists of 100 categories of crops. Chemical feature of
techniques that analysis information and perceive pattern, and dataset thus define the availability of nutrients and
that are used as classification and regression analysis. LS- micronutrients in soil. Availability of these nutrients and
SVM technique is suggested for large scale problem. LS-SVM micronutrients can be divided into different categories which
method is applied and it included various step. First step is to can be used to decide its effects on the yielding capability of
initiation and the tune the data. Then this method is used to crops. First in pre-processing, which all features are extract
train and test the input data. This algorithm is help to predict from input data. Then apply the classify techniques such as
the category of crop based on physical and chemical LS-SVM and deep neural network learning to classified the
characteristics of crop and soil. feature of plant and soil. Classifiers are used to analyse and
recognise pattern used for classification methods.
6
IJRITCC | August 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 8 05 - 07
______________________________________________________________________________________
The accuracy of the system is change using different data Journal of Research in Engineering and Technology, Volume:
set. Techniques have been run to train and accuracy is 04 Issue: 01, Jan-2015.
obtained according the different data set along prediction done [8] M.C.S.Geetha, “Implementation of Association Rule Mining for
different soil types in Agriculture”, International Journal of
to test the input data and give the name of crop, time duration Advanced Research in Computer and Communication
and pesticide name that effect the yield of crop. Engineering Vol. 4, Issue 4, April 2015.
A system contains two processes. The first process is used [9] Rashmi Arora, Savneet Kaur, “Mining of Soil Data Using
for generating a trained model. The algorithms are applied on Unsupervised Learning Technique”, International Journal of
the training dataset to train the model. The second process is Applied Engineering Research, ISSN 0973-4562 Vol. 7 No.11
used to test the already trained model, thus making predictions (2012).
and also measure the performance with respect to given [10] Mrs.Sagunthaladevi.S, Dr.Bhupathi Raju, Venkata Rama Raju
feature of crop and soil. “Survey on classification methods in data mining”, International
Journal of Computer Science Engineering (IJCSE), Vol. 5 No.03
May 2016.
TABLE I. PHYSICAL AND CHEMICAL FEATURE OF CROP AND
SOIL [11] G. Nasrin Fathima, R.Geetha, “Agriculture Crop Pattern Using
Data Mining Techniques” Volume 4, Issue 5, May 2014.
Physical feature of crop Chemical feature of soil
[12] Dr. S.Hari Ganesh, Mrs. Jayasudha, “Data Mining Technique to
Temperature, Rainfall, PH range, 5types of soil such as Predict the Accuracy of the Soil Fertility” IJCSMC, Vol. 4,
Humidity black soil, alluvial soil, red soil, Issue. 7, pg.330 – 333, July 2015.
laterite soil and desert soil, soil [13] Jyotshna Solanki, Prof. (Dr.) Yusuf Mulge, “Different
moisture, iron, aluminium, Techniques Used in Data Mining in Agriculture” , Volume 5,
magnesium, nitrogen, potash and Issue 5, May 2015.
phosphorous [14] GAO Yi-yang, Ren Nan-ping, “Data Mining And Analysis Of
Our Agriculture Based On The Decision Tree”, ISECS
TABLE II. COMPASION OF ACCURACY International Colloquium on Computing, Communication,
Control, and Management, (2009).
Sr. No. Classification Techniques Accuracy
[15] Raorane A.A., Kulkarni R.V., “Data Mining: An effective tool
1. LS-SVM 98.8% for yield estimation in the agricultural sector”, International
2. Deep neural network learning 95.53% Journal of Emerging Trends & Technology in Computer Science
(IJETTCS), Volume 1, Issue 2, July – August 2012.
I. CONCLUSION
Yield prediction is very complex problem in agriculture
field and also effect the economic growth of the farmer. In this
paper, we developed the system to achieve high accuracy and
increase the production of crop. We use the two techniques
such as LS-SVM and deep neural network learning to predict
the yield of crop based on available physical and chemical
feature of plant and soil and solves the yielding problem in
agriculture area. This research determines the name of crop,
time duration and pesticide name used in particular crop. This
system helps the farmer to take the right for sowing the crop in
available land.

REFERENCES
[1] B Vishnu Vardhan and D. Ramesh. “Density Based Clustering
Technique on Crop Yield Prediction”, International Journal of
Electronics and Electrical Engineering Vol. 2, No. 1, March,
2014.
[2] A. Mucherino, Petraq Papajorgji, P. M. Pardalos, “A survey of
data mining techniques applied to agriculture”, 2009.
[3] Monali Paul, Santosh K. Vishwakarma, Ashok Verma,
“Analysis of Soil Behaviour and Prediction of Crop Yield using
Data Mining Approach”, International Conference on
Computational Intelligence and Communication Networks,
2015.
[4] P.Bhargavi, Dr.S.Jyoth, “Applying Naive Bayes Data Mining
Technique for Classification of Agricultural Land Soils”,
IJCSNS International Journal of Computer Science and Network
Security, VOL.9 No.8, August 2009.
[5] D Ramesh , B Vishnu Vardhan, “Data Mining Techniques and
Applications to Agricultural Yield Data”, International Journal
of Advanced Research in Computer and Communication
Engineering Vol. 2, Issue 9, September 2013.
[6] Georg Ruß, Rudolf Kruse, Martin Schneider, and Peter Wagner,
“Data Mining with Neural Networks for Wheat Yield
Prediction”, 2008.
[7] D Ramesh , B Vishnu Vardhan, “Analysis of crop yield
prediction using data mining techniques” IJRET: International

7
IJRITCC | August 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________