Published by : International Journal of Engineering Research & Technology (IJERT)

http://www.ijert.org ISSN: 2278-0181

Vol. 9 Issue 08, August-2020

Crop Prediction using Machine Learning

Approaches
Nischitha K
Assistant Professor,
Department of E&C Engineering
P.E.S College of Engineering, Mandya India

Dhanush Vishwakarma Mahendra N

Department of E&C Engineering Department of E&C Engineering
P.E.S College of Engineering, P.E.S College of Engineering,
Mandya India. Mandya, India

Ashwini Manjuraju M.R

Department of E&C Engineering Department of E&C Engineering
P.E.S College of Engineering, P.E.S College of Engineering,
Mandya India. Mandya, India.

Abstract— As we know the fact that, India is the second designed the system using a machine learning for betterment
largest population country in the world and majority of people in of the farmer. Machine learning(ML) is a game changer for
India have agriculture as their occupation. Farmers are growing agriculture sector. Machine learning is the part of artificial
same crops repeatedly without trying new verity of crops and they intelligence, has emerged together with bigdata technologies
are applying fertilizers in random quantity without knowing the
deficient content and quantity. So, this is directly affecting on
and high-performance computing to create new opportunities
crop yield and also causes the soil acidification and damages the for data intensive science in the multi-disciplinary agri-
top layer. So, we have designed the system using machine learning technology domain. In the Agriculture field machine learning
algorithms for betterment of farmers. Our system will suggest the for instance is not a mysterious trick or magic, it is a set of well
best suitable crop for particular land based on content and define model that collect specific data and apply specific
weather parameters. And also, the system provides information algorithms to achieve expected results [7].
about the required content and quantity of fertilizers, required
seeds for cultivation. Hence by utilizing our system farmers can The designed system will recommend the most suitable crop
cultivate a new variety of crop, may increase in profit margin and for particular land. Based on weather parameter and soil
can avoid soil pollution.
content such as Rainfall, Temperature, Humidity and pH. They
Keywords—Machine Learning,Crop prediction,Decision tree,SVM, are collected from V C Farm Mandya, Government website
Rainfall prediction,Crop recommendation; and weather department. The system takes the required input
from the farmers or sensors such as Temperature, Humidity
I. INTRODUCTION and pH. This all inputs data applies to machine learning
predictive algorithms like Support Vector Machine (SVM) [5]
Agriculture is one of the important occupation practiced in and Decision tree [6] to identify the pattern among data and
India. It is the broadest economic sector and plays a most then process it as per input conditions. The system
important role in the overall development of the country. More recommends the crop for the farmer and also recommends the
than 60% of the land in the country is used for agriculture in amount of nutrients to be add for the predicted crop. The
order to suffice the needs of 1.3 billion people Thus adopting system has some other specification like displaying
new agriculture technologies is very important. This will be approximated yield in q/acre, required seed for cultivation in
leads the farmers of our country towards profit [1]. Prior crop kg/acre and the market price of the crop.
prediction and yield prediction was performed on the basis of
farmers experience on a particular location. They will prefer II. LITERATURE SURVE
the prior or neighborhood or more trend crop in the Ashwani kumar Kushwaha[2] describes crop yield prediction
surrounding region only for their land and they don’t have methods and a suggest suitable crop so that it will improve the
enough of knowledge about soil nutrients content such as profit for the farmer and quality of the agriculture sector. In
nitrogen, phosphorus, potassium in the land. Being this as the this paper for crop yield prediction they obtain large volume
current situation without the rotation of the crop and apply an data, it’s been called as big data (soil and weather data) using
inadequate amount of nutrients to soil it leads to reduce in the Hadoop platform and agro algorithm. Hence based repository
yield and soil pollution (soil acidification) and damages the top data will predict the suitability crop for particular condition
layer. Considering all these problems takes into the account we and improvement crop quality.

IJERTV9IS080029 www.ijert.org 23
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 9 Issue 08, August-2020

Girish L [3] describe the crop yield and rain fall prediction define the target for a model. After data cleaning the dataset
using a machine learning method. In this paper they gone will be split into training and test set by using sklearn library.
through a different machine learning approaches for the
prediction of rainfall and crop yield and also mention the 3.3. Machine Learning Algorithm for Prediction: -
efficiency of a different machine learning algorithm like liner Machine learning predictive algorithms has highly optimized
regression, SVM, KNN method and decision tree. In that estimation has to be likely outcome based on trained data.
algorithm they conclude that SVM have the highest efficiency Predictive analytics is the use of data, statistical algorithms and
for rainfall prediction. machine learning techniques to identify the likelihood of future
outcomes based on historical data. The goal is to go beyond
Rahul katarya [4] describes the different machine learning knowing what has happened to providing a best assessment of
methods used for accelerating crop yield. In this paper they what will happen in the future.
gone through different artificial intelligence techniques such In our system we used supervised machine learning algorithm
as machine learning algorithm, big data analysis for precision having subcategories as classification and regression.
agriculture. They explain about crop recommender system Classification algorithm will be most suitable for our system.
using KNN, Ensemble-based Models, Neural networks, ...etc.
➢ Rainfall prediction: -SVM algorithm.
➢ Crop prediction: - Decision tree algorithm.
III. PROPOSED SYSTEM
A. Rainfall Prediction: The flow chart of rainfall prediction
The Proposed system will predict the most suitable crop for
as shown in fig 2
particular land based on soil contents and weather parameters
such as Temperature, Humidity, soil PH and Rainfall.
START

GOVTWEBSITES Display Suitable CROP

AND VC FORM
Load external dataset
MANDYA
MACHINE LEARNING Display Required NPK
nutrients for crop
DATA PREPROCESSING ALGORITHM FOR CROP
AND RAINFALL CROP RECOMMENDATION
DATA COLLECTION (Data cleaning, dealing DATA PREPROCESSING
with missing values, PREDICTION Display Required Seeds
(Combining data from in Kg/acre for cultivation (dealing with missing values, data cleaning ,train/test
CLIMATE DATA train/test split) (SVM, Decision Tree
different resources) split)
Algorithm )
Provide information about
current market price and
approximate yield of
recommended crop

AGRICUTURE Predict Rainfall using SVM algorithm

MARKETING
INFORMATION

Figure 1. Architecture of the proposed system.

The Architecture of the proposed system consists of various Print predicted rainfall for 2020

blocks as shown in the fig (1) as follows

3.1. Data Collection: - Data collection is the most efficient END

method for collecting and measure the data from different
resources like govt websites, VC Form Mandya, APMC Figure 2. Flow chart for rainfall prediction.
website .... etc. To get an approximate dataset for the system.
This dataset must contain the following attributes The process starting with a load the external dataset contains
i)Soil PH ii) Temperature iii) Humidity iv) Rainfall v) Crop previous year rainfall data. Then loaded dataset pre-processing
data vi) NPK values, those parameters will consider for crop takes place as discussed in Data Pre-processing section. After
prediction. For the annual rainfall prediction, we collect data pre-processing we train the model by using SVM
previous year rainfall data. classifier with Radial basis function(RBF) kernel. Then fitting
the classifier into training set. Radial basis function can be
3.2. Data Preprocessing: - After collecting datasets from written in the mathematical form in equation (1),
various resources. Dataset must be preprocessing before
2
training to the model. The data preprocessing can be done by 𝐾(𝑋1, 𝑋2) = 𝑒𝑥𝑝𝑜𝑛𝑒𝑛𝑡 (−𝛾||𝑋1 − 𝑋2|| ) …. ..(1)
various stages, begins with reading the collected dataset the
process continues to data cleaning. In data cleaning the
Where, ||X1-X2|| = Eucliden distance between X1 & X2
datasets contain some redundant attributes, those attributes are
𝛾 = Gamma
not considering for crop prediction. So, we have to drop
After fitting and test the model, at last this model will predict
unwanted attributes and datasets containing some missing
the appropriate annual rainfall. The predicted rainfall is the one
values we need to drop these missing values or fill with
of the input parameter for the crop prediction system.
unwanted nan values in order to get better accuracy. Then

IJERTV9IS080029 www.ijert.org 24
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 9 Issue 08, August-2020

B. Crop Prediction: The overall output is shown with Graphical user interface as
Crop prediction process being with the loading the external shown in figure (4),
crop datasets. Once the dataset read then pre-processing will
be done by various stages as discussed in Data Pre-processing
section. After the data pre-processing, train the models using
Decision tree classifier into training set. For a prediction of the
crop, we consider a various factor such as temperature,
humidity, soil PH and predicted rainfall. Those are the input
parameter for a system that can be entered by manually or
taken from the sensors. Predicted rainfall and input parameter
values will be appended in a list. The Decision tree algorithm
will predict the crop based on list data.
The overall flow chart of the proposed system is shown in the
figure (3).
START

Load External Crop dataset Load Rainfall dataset

DATA PREPROCESSING
(dealing with missing values,
Figure 4. Overall output with GUI
DATA PREPROCESSING
(dealing with missing values, data cleaning data cleaning ,train/test split)
,train/test split)
Predict Rainfall using SVM NO

Read temperature,
algorithm We tested the system for various data set that has been
Humidity, pH Values
collected from the different farmers for their lands conditions.
Append all the values in list YES Predict Rainfall for
2020
Land condition’s just like the land having different pH,
YES humidity and NPK values. The predicted annual rainfall is
Predict Crop Using Decision Tree Regressor Print predicted Rainfall constant for year 2020.
Crop Classification using
NO Predicted crop index values

YES

Display Suitable Crop

Read NPK Values

Display Required NPK,

Required Seed, Yield and
Price

END

Figure 3. Proposed System Flow chart.

Figure 5. Tested output results
3.4. Crop Recommendation:
Based on predicted rainfall, soil contents and weather
parameters the system will recommend the most suitable crop V. CONCLUSION AND FUTURE SCOPE
for cultivation. This system also provides details about required CONCLUSION:
fertilizers like Nitrogen(N), Phosphorus (P) and potassium(K) Presently our farmers are not effectively using technology and
in Kg per hectare and display the required seed for a cultivation analysis, so there may be a chance of wrong selection of crop
in Kg per acre for recommended crop. for cultivation that will reduce their income. To reduce those
This system as contain some other feature such as display the type of loses we have developed a farmer friendly system with
current market price and approximated yield in quintal per acre GUI, that will predict which would be the best suitable crop
for recommended crop. Those all details will helps to farmers for particular land and this system will also provide
for choosing the most profitable crop. information about required nutrients to add up, required seeds
for cultivation, expected yield and market price. So, this makes
IV. EXPERIMENTAL OUTCOME the farmers to take right decision in selecting the crop for
cultivation such that agricultural sector will be developed by
The proposed system recommends the best suitable crop for innovative idea.
particular land by considering parameters as annual rainfall,
temperature, humidity and soil pH. Among these parameters FUTURE SCOPE:
annual rainfall is predicted by system itself by using previous We have to collect all required data by giving GPS locations
year data with SVM algorithm and other parameters are have of a land and by taking access from Rain forecasting system of
to be entered by the user. In the output section the system by the government, we can predict crops by just giving GPS
displays a suitable crop, required seeds/acre, market price and location. Also, we can develop the model to avoid over and
approximate yield of the recommended crop and also the under crisis of the food.
system takes NPK values in the input section to display the
required NPK for the recommended crop.

IJERTV9IS080029 www.ijert.org 25
(This work is licensed under a Creative Commons Attribution 4.0 International License.)
Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 9 Issue 08, August-2020

REFERENCES [4] Rahul Katarya, Ashutosh Raturi, Abhinav Mehndiratta, Abhinav

Thapper “Impact of Machine Learning Techniques in Precision
[1] Prof. D.S. Zingade ,Omkar Buchade ,Nilesh Mehta ,Shubham Agriculture”.
Ghodekar ,Chandan Mehta “Crop Prediction System using Machine [5] Pijush Samui, Venkata Ravibabu Mandla, Arun Krishna and Tarun
Learning”. Teja “Prediction of Rainfall Using Support Vector Machine and
[2] Ashwani kumar Kushwaha, Swetabhattachrya “crop yield prediction Relevance Vector Machine”.
using agro algorithm in hatoop”. [6] Himani Sharma, Sunil Kumar “A Survey on Decision Tree
[3] Girish L, Gangadhar S, Bharath T R, Balaji K S, Abhishek K T “Crop Algorithms of Classification in Data Mining”.
Yield and Rainfall Prediction in Tumakuru District using Machine [7] Pavan Patil, Virendra Panpatil, Prof. Shrikant Kokate “Crop
Learning”. Prediction System using Machine Learning Algorithms”.

IJERTV9IS080029 www.ijert.org 26
(This work is licensed under a Creative Commons Attribution 4.0 International License.)