RECOMMENDATION OF CROP AND PESTICIDES USING MACHINE LEARNING

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1352
RECOMMENDATION OF CROP AND PESTICIDES USING MACHINE
LEARNING
J. Ignashya preetha1, N. Priyadharshini2, P. Mageshwari3, S. Rakshana4, Dr. S. Jeyalakshmi5
1,2,3,4,5 Department of IT, SRM Valliammai Engineering College, Tamil Nadu 603203
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract – The major resource for improving the economy
of India is agriculture. From past farmers followed ancestral
faming pattern and regularities within it. A single farmer
cannot take action upon improving the crop yield of a nation
and does not have enough potential tomaximizethecropyield
by adopting technical norms within plant growth and
improving the yield in a large quantity. Severe change in
climatic condition and several other pesticides attack cause
shorting of crop yield and also led to food shortage. A simple
misguided decision in farming can affect a farmer severe. In
recent, there is lot of techniques applied by researchers and
those techniques are available to raise the quantity of yield.
This in turn changed traditional farming approach and
introduced precision farming. Recently data mining performs
vital role in identifying plant disease and providing solution
prescribing pesticides to plant disease. But this study extends
the application of data mining in agriculture to a greater
extent. The cultivation of precious crop at right time is the
major issues faced by farmer. This study proposes machine
learning (ML) approach to resolve it and makes the farmer to
choose right crop based on the nutritioncontentand qualityof
soil. The machine learning algorithmschosenforthisstudy are
Random forest, decision tree and K-nearestneighboring. Some
of the factors mainly considered for recommendation of plant
are humidity, rainfall, pH value, soil moisture. The
recommended technique makes farmer to take decision on
improving the crop yield; recommending crops as perclimatic
condition and quality of land.
Key Words: Agriculture, Crop Recommendation, Machine
Learning (ML), Random Forest, K-nearest Neighboring
(KNN), Decision Tree
1. INTRODUCTION
Agriculture is said to be the backboneofIndian economyand
it utilizing 60% of nation land to fulfil the food needs around
1.2 million people. Farmer doesn’t have conquered
knowledge about severe climatic changes and the soil
moisture content. Mostly famers are difficult to understand
those two factors. This in turn led to decrease in expected
level of productivity. The selection of pesticide, usage of
water and maintaining of it will make the crop growth even
stronger. Every crop has special climatic factors. By
precision farmingtechnique, thosefactorsarehandledasper
the crop planted. Precision farming not only focuses on
productivity but also raising the yield rate of crop. To make
agriculture as a profitable business for farmers and satisfies
the need of a nation, different kind of agricultural practices
are carried out. In developing nation like India, sustainable
agriculture is practised to manage the necessityoffood. Alot
of techniques were carried out to minimizetheshorteningof
crop yield; but traditional agriculture having its own
demerits. The demerits arefurtherlimitedthroughprecision
farming. Other than that, some other factors affecting the
yield of a plant are bacterial, fungal and viral diseases. The
detailed explanation of various plant diseases occurring
repeatedly in farms are given below:
Anthracnose: Mostly fungus is observed in genus
collectotrichum and other regions; lesions occur on stem.
The major reason for this disease is rotted waste andcertain
other wastages around it. During winter, the plants are
affected by this disease and it is transmitted to nearest plant
through wateringand pollination. Thedeadtissuesappearas
anthracnose.
Bacterial blight: Lesions are converted into dead spots in
this; later elongated lesions are appeared as like linear
streaks and it is turned into milky green colour. As like
anthracnose, this disease will affect in winter season and
transmission through insects and water.
Alternaria alternate: A fungal disease found in different
kinds of plants and the symptoms are observed as blights,
leaf spot and rots. The spores in a leaf are createdbyconidia.
Rainfall and humidity are the comfort zone for this disease.
Cercospora leaf spot: This does not have sexual stage and
its genus is mycospharella.
Later data mining and ML techniques are used by
researchers to bring revolution in the field of traditional
agriculture to maximize the productivity by considering the
necessity. ML can gain expertise without doing additional
programming in a machine, so it maximizes the
accomplishment of machinebydifferentiatinganddepict the
consistency and format of drive data. In this research,
combination of three different algorithms such as Random
forest, KNN, decision tree algorithms were used to suggest
crop, fertilizer and pesticides. As per the land condition, the
proposed study will recommend crops and several other
essentialities. This type of recommendation is carried out
with the consideration of water level, moisture content, pH,
temperature.

From a selected region, the data are collected from soil
testing laboratory as well as from data.World. A wide
variety of data is chosen for this analysis and has been
processed through suggested ML approach. Following this
crops are suggested and requiredfertilizerischosen;disease
is identified and pesticides are selected. The aim of the
proposed study is to plant crops at right time and maximize
the yield.
2. RELATED WORKS
K.Venkataramana et.al [1], proposed a Ml approach such as
id3 algorithm to improving yield of tomato. The factors such
as csv, moisture level, temperature are considered in
selected dataset and the examination is done in Php
platform.
R. Sujatha et.al, developed a data mining approach to
improvise the yield of crop in a large cultivable land. The
parameters mainly considered for this study are name of
crop, type of soil, weather condition, seedlingselection,level
of water, pH value. The commonly affecting disease and the
disease which are ease to affect are considered secondary
during this process [2].
A. K. Tripathy et.al in [3], recommends Support Vector
Mechanism (SVM) classifier algorithm to find out suitable
crop for selected cultivable land by considering location, air
moisture level, seed varieties. The dataset is prepared
through Weka tool which makes pair of rules on present
dataset. Through python, the entire process is done.
S. Veenadhari, B. Misra and C. Singh, created a website to
know climatic condition of an area and yield of crop by c4.5
algorithm known as Crop Advisor. Dependent on c4.5
algorithm, decision tree and rule have been built. This
explains how the crop is affected by the change in climatic
condition [4].
Jun Wu et.al, suggested varieties of crops which is able to
accept and grow under variable climatic condition. For this
approach, decision tree classifier algorithm is utilized and it
utilizes new factors which was helpful to improvisetheyield
of crops. 10-fold cross validation method is used to check
dataset, horse-colic and soyabean dataset [5].
Murali Krishna et.al, defines interfacing of data mining
technique with humidity and pesticide attack onplants.This
explains the difficulties faced by farmers and problems in
interfacing data miningtechniquewithagriculture. Pesticide
attack prevention is done by recommending pesticides [6].
Verheyen et.al, in [7], described about statistical mining
approach to review the characteristics of soil. The K-means
clustering classifier approach is utilized to classifying soil
type and this system combines with GPS to select a region
and does categorization.
Radhike et.al in [8] suggested smart farming approach by
utilizing sensing devices. To calculate the intensity of light,
yhe proposed technique utilizes pH, temperature and
moisture sensor. The sensors are placed to sense the
environmental condition and it collects the information of
environment which is delivered towardsprocessingwithout
any change within it. This is suited for all kind of crop
recommendation as per the environmental factors.
Kumar et.al, proposed an IoT model to guide farmers to
select suitable crop and fertilizers. This system uses pH
sensor and soil moisture sensor which is connected with
radio frequency system to transmit data. The received
information is processed by a ML approach so called
decision tree. The necessary information suchasselectionof
fertilizer, time of providing fertilizers tocropsandcultivable
stage is send to farmer. Through android application, the
farmer will receive guidelines.
Suhas Athani et.al proposed neural network to process the
data delivered by IoT based system. The neural network
helps to find the condition of cultivable land in association
with sensing devices and the farmer is guided through
android application [10].
3. METHODOLOGY
The purpose of this proposed system is to go over the best
practices for predicting crop area and yield in mixed and
continuous cropping systems. List and area frames are both
taken into consideration. The sampling frame for crop yield
estimation will be determined by the sample selected for
area estimation. It is briefly covered the samplingprocedure
used to select the sample for crop area and yield calculation.
To merge the subjective and objective techniques, a double
sampling regression estimator is used. The domain
estimation technique and a double sampling regression
estimator are also used to estimate crop area and yield. The
notion of domain estimation enables the estimation of crop
area and yield for a variety of crops and blends from a single
sample. The criterion for estimating sample size is also
included.
Fig -1: Architecture diagram

The detailed explanation of eachmoduleisexplainedbellow.
Collection of dataset: The dataset is made up of soil-
specific properties that are gathered and evaluated in a soil
testing lab. Moreover, similar web sources of general crop
data were utilised. Rice, maize, chickpea, kidneybeans,
pigeonpeas, mothbeans, mungbean, blackgram, lentil, and
other crops are included in our model.Inthetrainingdataset,
the number of instances of each crop is shown. Ph,
temperature,rainfall,andhumiditywereallfactorstakeninto
account.
Data cleaning: Cleaning data is an essential part of every
machinelearning research. Datacleaningisperformedinthis
module to prepare data for analysis by removingorchanging
data that is erroneous, incomplete, redundant, or badly
formatted. In tabular data, you can investigate your data
using a variety of statistical analysis and data visualisation
approaches to find data cleaning activities you might wish to
do.
Fig -2: Flow analysis of proposed study
Training and testing of datasets: Separating the
attributes and assigning the variables as X and Y as train and
test data based on the dependent and independentvariables.
Feature extraction: This is done to decreasethecountof
attributes in the dataset, resulting in benefits such as faster
training and improved accuracy.
The featureis the majorsourceforcomputationalmethod
of solving problems and tasks. Some image structure like as
points, edges are characteristic. The extraction of features
dataset is utilized for classification. Every character in the
feature extraction is denoted by an identification feature
vector. The main objective of this stage is to extract a
collection of features that improvise prediction ratewithfew
element count and createsamefeatureforvariousinstanceof
the same symbol.
Classifier Algorithms:
The most popular ML approach used here is Random
forest also called as ensemble learning which a supervised
learning technique is. It is mainly used as classifier and
solving complex problems. It utilizes multiple numbers of
classifiers to perform classification and maximize the
performance in an effective manner. This technique have
decision tree in a large count. To raise the accuracy of the
model this method is suited. The prediction process is done
with the usage of each tree without depending on a decision
tree. Depending on the majority votes of prediction, this
algorithm finalizes the output. With high dimensionality this
method process large volume of dataset. Also it avoids over
fitting issue to maximize the accuracy as much as possible.
KNN is a data mining technique. It takes every
characteristic in training set as various dimensions in some
space, and take the value an observation has for this
characteristic to be its coordinate in that dimension, so
getting a set of points in space. We can then consider the
similarity of two points to be the distance among theminthis
space under some appropriate metric. The way in which the
algorithm decides which of the points from the training set
are similar enough to be considered when choosing the class
to predict for a new observation is to pick the k closest data
points to the new observation, and to take the most common
class among these.
Decision Tree (DT) is a foreboding representation which
functions by testing states at each stage of tree and tends to
reach end tree in between that multiple decisions are
recorded. The state depends on the application and the
output is form in the form of decision. This algorithm
calculates information gain of essential attributes such as
area, vaporpressure, yieldand cloudcover.Thetwodifferent
much needed attribute suchas areaandyieldisinconnection
to make a decision and then extended.
4. RESULT AND DISCUSSION
The Random forest technique attained greater efficiencyand
it is visualized in chart 1. The decision tree attained 78%
accuracy, KNN have 83% and finally Random forest attained
97%. The experimental analysis is more effective and can
suggest right crop at right time to raise the yield.

Chart -1: Accuracy Comparison
To recommend fertilizer the attributes considered are
humidity, moisture level, soil type and finally crop to be
grown. Fertilizer recommendation follows loading of
external fertilizer datasets. The values of attributes are
chosen either by sensors or human records.
5. CONCLUSIONS
In developing countries like India agriculture plays a vital
role. Farmer and the country economy will rise when
production of agricultural product is at larger proportion.
The proposed system examines the quality of land,
recommend pesticides and fertilizers. By this proposed
study, farmer can planting suitable crops and can get more
yields. This in turn led to improving the economy and there
is no doubt within it. The experimental analysis evident that
Random forest based crop selection would be a right choice.
In future, the proposed study led to a greater extends by
considering additional features from dataset and doublethe
productivity.
REFERENCES
[1] CH. Vishnu Vardhanchowdary, Dr.K.Venkataramana,
“Tomato Crop Yield Prediction using ID3”, March
2018,IJIRT Volume 4 Issue 10 pp,663-62.
[2] R. Sujatha and P. Isakki, “A study on crop yield
forecasting using classification techniques” 2016
International Conference on Computing Technologies
and Intelligent Data Engineering (ICCTIDE'16),
Kovilpatti, 2016, pp. 1-4.
[3] N. Gandhi, L. J. Armstrong, O. Petkar and A. K. Tripathy,
“Rice crop yield prediction in India using supportvector
machines” 2016 13th International Joint Conference on
Computer Science and Software Engineering (JCSSE),
KhonKaen, 2016, pp. 1-5.
[4] S. Veenadhari, B. Misra and C. Singh, “Machine learning
approach for forecasting crop yield based on climatic
parameters” 2014 International Conference on
Computer CommunicationandInformatics,Coimbatore,
2014, pp. 1-5..
[5] Jun Wu, AnastasiyaOlesnikova, Chi-Hwa Song,WonDon
Lee (2009), “The Development and Application of
Decision Tree for Agriculture Data” IITSI, pp 16-20
[6] KiranMai,C., Murali Krishna, I.V, an A.VenugopalReddy,
“Data Mining of Geospatial Database for Agriculture
Related Application”, Proceedings of Map India,New
Delhi, 2006,pp 83-96.
[7] Verheyen, K., Adrianens, M. Hermy and S.Deckers
(2001),“High resolution continuous soil classification
using morphological soil profiledescriptions”Geoderma,
101:31-48.
[8] Radhika, Y., & Shashi, M. (2009). Atmospheric
Temperature PredictionusingSupportVectorMachines.
International Journal of Computer Theory and
Engineering, 55–58.
[9] Kumar, R., Singh, M. P., Kumar, P., & Singh, J. P. (2015).
Crop Selection Method to maximizecrop yieldrateusing
machine learning technique. IEEE Xplore.
[10] Athani, S., Tejeshwar, C. H., Patil, M. M., Patil, P., &
Kulkarni, R. (2017, February 1). Soil moisture
monitoring using IoT enabled arduino sensors with
neural networks for improving soil management for
farmers and predict seasonal rainfall forplanningfuture
harvest in North Karnataka — India

RECOMMENDATION OF CROP AND PESTICIDES USING MACHINE LEARNING

More Related Content

Similar to RECOMMENDATION OF CROP AND PESTICIDES USING MACHINE LEARNING (20)

More from IRJET Journal (20)

Recently uploaded (20)

RECOMMENDATION OF CROP AND PESTICIDES USING MACHINE LEARNING