A Review On Data Mining Techniques For Fertilizer Recommendation
A Review On Data Mining Techniques For Fertilizer Recommendation
ABSTRACT
Agriculture plays a crucial role in the life of an economy. It is the backbone for developing countries like India
as more than 70% of population depends on agriculture. To increase crop production many factors are
responsible like soil, weather, rain, fertilizers and pesticides. We have used soil parameters to increase crop
production because it is an essential key factor of agriculture. To maintain nutrient levels in the soil in case of
deficiency, fertilizers are added to soil. The common problem existing among the Indian farmers is that they
choose approximate amount of fertilizers and add them manually. Excess or insufficient addition of fertilizer
can harm the plant life and reduce the yield. This paper provides review of various data mining techniques used
on agriculture soil dataset for fertilizer recommendation. Mainly I focused on various soil parameters like Fe, S,
Zn, Cu, N and Ph value etc. In this survey, we also describe some Agriculture problems that can be solved by
using data mining techniques.
Keywords : Agriculture, Soil Fertility, Fertilizer Recommendation, Data Mining, Clustering, Classification,
Neural Network
CSEIT1831400 | Received : 16 Feb 2018 | Accepted : 28 Feb 2018 | January-February-2018 [(3) 1 : 1386-1390 ] 1386
help of data mining concepts and they can get transaction. For example, this technique is used in
optimized results from recommender. [23]. The market basket analysis, shopping cart analysis,
motivation behind this work is to explore data online advertisement etc [2] [6].
mining techniques, which are suitable for solving In this method, the focus is on finding relationships
complex agricultural problems. between the different items in a transactional
database. For example, if there is set of items as I=
Researchers design and develop applications for {i1,i2, ……., in} and the set of transaction as D= {t1,
solving complex agricultural problems using data t2, …..tn} then the association rule could be,
mining like Yield Data Prediction, Soil Mapping and X => Y, Where X and Y belongs to set I. [19]
Classification, Fertilizer and Pesticide Management, B. Clustering
Grading and sorting of agriculture product etc [24]. Clustering means identification of common kind of
objects. By using clustering method user can
II. SOIL ANALYSIS FOR FERTILIZER additional identify dense and sparse regions in
RECOMMENDATION object space and can find overall division pattern
among data attributes.
Among various trust areas of agriculture research as
crop cultivation, market analysis and price prediction,
rainfall prediction, classification of soil plays an
important role because analyzing soil provides major
contribution to the support of the farmers. Data
mining in agricultural soil datasets is exciting and
modern research area [19].
The productive capacity of a soil depends on soil
fertility. Soil fertility is the capacity of the soil to
support the growth of plants and it is measured by
Figure 1: Clustering
the amount of macro and micronutrients, water, pH
Classification way can also be used for productive
etc [10].To maintain nutrient levels in the soil in case
means of Distinguishing category or category of
of deficiency, fertilizers are added to soil. Better crop
object but it becomes costly so clustering can be
yield is highly depends on inputs of fertilizer.
used preprocessing way for attribute subset choice
Fertilizer not only increases crop yields, but also
and categorization [2] [3].
builds soil organic matter [24].
C. Classification
This paper provides analysis of different data mining
Classification is a Technique which forms the
techniques to analyze the soil test data sets for
common stepping stone for various recursive
fertilizer recommendation.
algorithms and methods designed for data mining.
The design aim of a classification algorithm is to
III. TECHNIQUES USED IN DATA MINING
divide various data-sets or database which have
enough data into different types of sets so that they
There are Various Techniques available for data
can be analyzed using a single step exhaustive model.
mining which can also be used for agriculture
Classification method makes use of statistical
domain.
techniques like decision trees, linear programming,
A. Association rule mining:
neural network and statistic-oriented approach, etc.
Association is a data mining technique in which a [6]
pattern is discovered based on a relationship of a
particular item on other items in the same
Volume 3, Issue 1, January-February-2018 | www.ijsrcseit.com | UGC Approved Journal [ Journal No : 64718 ] 1387
D. Regression: D. SVMs in Agriculture: Support vector machine
Regression is a one type of data mining Method used (SVM) which was originally developed by Vapnik
to fit a formula to a dataset. The process for (1998) has been widely applied to many different
estimating the relationships among variables is fields, such as signal process and time series analysis.
known as regression analysis. Regression analysis SVMs are one of the newest supervised machine
can be helpful to understand how the typical value learning techniques. The current study investigated
of the dependent variable can be changed. [5][6] the applicability of support vector machine in
agriculture is in the crop classification and in the
IV. THE SCOPE OF DATA MINING IN analysis of the climate change scenarios. [21][18]
AGRICULTURE E. Decision Tree in Agriculture: Decision Tree is
tree-shaped structures that represent sets of
There are some research area of agriculture where decisions and generate rules for the classification of
data mining is useful in many ways. It may provide a dataset. Specific decision tree methods include
crucial role in decision making for complex Classification and Regression Trees (CART) and Chi
agricultural problems. Some research challenges of Square Automatic Interaction Detection (CHAID).
agriculture can be solve by data mining algorithm as In agriculture, Decision tree algorithm is used for
follow: predicting soil fertility. [2][21]
A. K-mean in Agriculture: The K-means algorithm is
distance based clustering techniques. By applying V. RELATED WORK
this algorithm, K cluster are formed. Based on
Euclidean distance, object is placed into the P. Jasmine Sheela. Sivaranjani [20] made a comparative
respective cluster. [19] The k-means algorithm is used study of various classification algorithms such as
to classify soil in combination with GPS. Random forest, J48, Naïve Bayes and concludes that
Classification of plant and soil, grading apples before accuracy level for J48 algorithm is 98.17%, for
marketing, Monitoring water quality change, Random forest algorithm is 90.81%, for Naïve Bayes
Detecting weeds in precision agriculture, and the algorithm is 77.18%, for JRip algorithm is 92.53%,
prediction of wine fermentation problems can be and after using boosting for J48 it increase the
performed by using a k-means approach. [21]
accuracy level as 96.73%. This algorithm is used to
B. K-nearest neighbor in Agriculture: In pattern analyze a large number of datasets and it is used to
recognition, the k-Nearest Neighbors algorithm is a create a soil fertility prediction in easy manner.
non-parametric method used for classification and Jay Gholap [7] used Decision Tree algorithm for
regression. In Agriculture, k-NN algorithm is used in predicting soil fertility. He also used Selection and
simulating daily precipitations and other weather Boosting techniques to tune the performance of J48
variables and Estimating soil water parameters and algorithm. Ad boost is one of the boosting techniques
Climate forecasting. [21]
used in this research. After using selection and
C. Neural Networks in Agriculture: In data mining a boosting algorithm it increase the accuracy level to
statistical model known as Artificial Neural Network 96.73%.
is a Non-linear predictive model that learns through S.S.Baskar, L.Arockiam, S.Charles [8] classifies the soil
training and resembles biological neural networks in dataset using various Classification algorithms such
structure. [2] The neural network is used in Prediction as regression, J48, Naive Bayes. J48 gives the best
of flowering and maturity dates of soybean and in result and accuracy level is 93.86%. For given soil
forecasting of water resources variables in samples this system will recommend the fertilizer
agriculture. appropriately.
Volume 3, Issue 1, January-February-2018 | www.ijsrcseit.com | UGC Approved Journal [ Journal No : 64718 ] 1388
Vrushali Bhuyar [9] focuses on classification of soil Scholar for getting an efficient algorithm selection in
sample of Aurangabad District to predict fertility rate. according to the fertilizer recommended based on
She use J48, Naïve Bayes, and Random forest soil fertility.
algorithm to calculate fertility rate of soil data. Her
Study shows that among the classifier J48 classifier
perform best to predict fertility index. This will help
VII. REFERENCES
to decision maker to recommend fertilizers
accordingly.
[1]. https://en.wikipedia.org/wiki/Data_mining
[2]. Neha Patel, Rakesh Patel, Shruti Upadhyay, “A
Amrutha A, Lekha R, A Sreedevi [10] says that the
Survey Paper of Study about Data Mining”
time taken for soil testing by chemical methods in a
International Journal for Research in Computer
laboratory takes a few days, while in the proposed
Science,2016
system the results are obtained within 30 minutes.
[3]. Simmi Bagga, Dr. G. N. Singh, “Applications of
The results obtained from the soil test are fed to
Data mining” ,international journal for Science
sensors and the results are analyzed using a
and Emerging Technology with latest Trends,
microcontroller which in turn needs a few seconds.
2012
The system has been developed for the controlled
[4]. https://www.tutorialspoint.com/data_mining/d
addition of fertilizers in order to avoid excess/
m_classification_prediction.htm
deficient fertilizers in the soil.
[5]. J. Han and M. Kamber, jianpei “Data mining:
Navneet, Nasib Singh Gill [12] propose a technique
Concepts and techniques (3rd edition)”,
that produces a compact decision tree having
Morgan Kaufman Publishes, 2012
increased classification accuracy and also Schwartz
[6]. Aashay kumar Dubey, Saurabh Kamath, Dhruv
criterion is used to get the optimal number of
Kanakia “Learning Data Mining Techniques”
clusters. The experimental results show that
International Journal of Computer
incorporating both clustering and classification
Applications, 2016.
algorithm is fast and more accurate for
[7]. Jay Gholap, “Performance tuning of j48
recommendation of fertilizers of real world soil
Algorithm for prediction of soil Fertility”,
dataset.
AJCSIT, 2012.
[8]. S. S. Baskar, L. Arockiam, S.Charles “Applying
VI. CONCLUSION
Data Mining Techniques on Soil Fertility
Prediction”, IJCAT, 2013.
There are a lot of opportunities for research on data
[9]. Vrushali Bhuyar, “Comparative Analysis of
mining in agriculture to develop practical applications
Classification Techniques on Soil Data to
to solve difficult problems in agriculture sector. This
Predict Fertility Rate for Aurangabad District”,
paper provides an analysis of data mining techniques
IJETTCS, 2014.
used in agriculture soil data and also compares the
[10]. Amrutha A, Lekha R, A Sreedevi, “Automatic
J48, Naïve Bayes, JRip, K-Means Classifier algorithm.
Soil Nutrient Detection and Fertilizer
Also suggest that an Artificial Neural Network is best
Dispensary System”, IEEE ,2016
other than other classification algorithms in data
[11]. Prof. NileshPatil, Apoorav Singh, Piyush Patel,
mining. It concludes that decision tree algorithm
Nisarg Joshi, “Soil Classification support for
performs best to analyze soil fertility condition and
Farming Decisions”, IJRITCC, 2017.
the results and finding motivate us to do further
[12]. Navneet, Nasib Singh Gill , “Algorithm for
research work on soil data for fertilizer
Producing Compact Decision Trees for
recommendation. This survey helps to the Research
Volume 3, Issue 1, January-February-2018 | www.ijsrcseit.com | UGC Approved Journal [ Journal No : 64718 ] 1389
Enhancing Classification Accuracy in Fertilizer
Recommendation of Soil”,IJCA,2014
[13]. Navneet and Nasib Singh Gill, “Classification
Using the Compact Rule Generation”, OJCST,
2015.
[14]. Jharna Majumdar, Sneha Naraseeyappa and
Shilpa Ankalaki, “Analysis of agriculture data
using data mining techniques: application of
big data”,Springer,2017
[15]. http://mpkrishi.org/EngDocs/AgriLeft/Soilt/soil
tHome.aspx#SS
[16]. Soil Test Interpretations and Recommendations
Handbook, Daryl D. Buchholz
[17]. A. Mucherino, Petraq Papajorgji, P. M.
Pardalos,” A survey of data mining techniques
applied to agriculture”, Springer, 2009
[18]. Niketa Gandhi, Leisa J. Armstrong, “A review
of the application of data mining techniques for
decision making in agriculture ”,IEEE,2016
[19]. N. Hemageetha,” A Survey on Application of
data mining techniques to analyze the soil for
agriculture purpose”, IEEE,2016
[20]. P. Jasmine Sheela, M.Sc., K. Sivaranjani, M.Sc.,
M.Phil., “A Brief Survey of Classification
Techniques Applied To Soil Fertility
Prediction”, ICETSH-2015
[21]. Hetal Patel, Dharmendra Patel,” A Brief
survey of Data Mining Techniques Applied to
Agricultural Data”, IJCA, 2014
[22]. http://bigdata-madesimple.com/14-useful-
applications-of-data-mining/
[23]. Kiran Shinde, Jerrin Andrei, Amey Oke,” Web
Based Recommendation System for Farmers”,
IJRITCC,2015
[24]. Tejas S. Mehta, Dr. Dhaval R. Kathiriya,
“Survey of Data Mining Techniques in
Precision Agriculture”, IJSR, 2015
Volume 3, Issue 1, January-February-2018 | www.ijsrcseit.com | UGC Approved Journal [ Journal No : 64718 ] 1390