JOURNAL OF ALGEBRAIC STATISTICS

Volume 13, No. 1, 2022, p. 426-430

https://publishoa.com
ISSN: 1309-3452

A Machine Learning Based Crop Recommendation System: A Survey

Rohini Jadhav1, Dr. Pawan Bhaladhare2

Research Scholar, School of Computer Science & Engineering, Sandip University1
Professor, School of Computer Science & Engineering2
rohinijadhav333@gmail.com1
pawan.bhaladhare@sandipuniversity.edu.in2

Abstract – Agriculture and related industries are the most important sources of employment and income in rural Areas.
The agricultural industry in the country also makes a significant contribution. Productivity as a percentage of GDP
(GDP). The agricultural industry is a huge source of wealth for the country. However, when compared to other
agricultural products, the yield per hectare is disappointing. Standards that are accepted all over the world. There are
numerous reasons why marginal farmers in India have a higher suicide rate. This is a study paper. Farmers are
recommended an effective and user-friendly yield prediction system. The proposed system connects Farmers through a
smartphone app. GPS technology aid in user identification and location. The user specifies the area and type of soil in
which they want to work, and machine learning algorithms enable the selection of the most profitable user-selected crop
yield prediction or crop list. Support Vector Machine (SVM), Artificial Neural Network (ANN), Random Forest (RF),
Multivariate Linear Network (MLN), and a combination of regression and KNN are used to estimate crop yields. The
Random Forest produced the best results of the three. The rate of accuracy is 95%. Aside from that, the system
recommends the best options available. It's time to start using chemical fertilizer to boost output.

Keywords- Agriculture, Crop Recommendation, Machine Learning

INTRODUCTION- non-farming use [4][5]. In addition, 48% of farmers

said they don't want their next generation to instead of
In India, there is a wide range of both physical and taking care of their farm, the next generation should
cultural diversity. Nearly everyone in India is reliant on urban locations are where I wish to live. As to why the
agricultural and agriculture-related jobs. Agriculture is reason for this is because farmers are notoriously inept.
benefiting greatly from the Internet of Things (IoT). Crop selection [9] is an example of this picking a crop
Farmers from a variety of issues and allow them to that will not provide a significant amount of income
concentrate on other relevant careers [1]. I believe that planting at the incorrect season, specific soil, and so
precision farming is one of the greatest innovations on. It's possible that the farmer bought the property
ever made. When it comes to illness prevention and from those who have never made a choice like this
early disease detection, it's doing all it can loss and before Possibly taken away from me. If you choose the
crop recommendations, as well as information on wrong crop, you’ll result in a lower return on
weather conditions, are supplied in addition to this. investment If the family is in good health, they will be
Exporting his goods and assisting in the preservation of happy [8]. if you rely on this revenue, it's quite tough to
the field His actions are controlled by sensors and make ends meet. Correct and timely information are
actuators tasks like irrigation and pesticide application both readily available and easily accessible. The lack of
should be scaled appropriately. By using all of these access to current knowledge deters would-be
methods, he may maximise his profits. Keep a close researchers due to my experience with case studies
eye on his area of expertise [3]. focusing on poor countries [6][7]. A mechanism has
Over half of the people in this country make their been put in place using the resources available to us.
living via agriculture [14,15]. According to the Presented a solution to this issue that will provide
Economic Survey for 2016-17, Farmers in 17 states insights into the long-term viability of crops machine
earn an average monthly wage of $1,050. Farmer learning models are used to make suggestions taught
suicides and diversion of Rs. 1700/-of farmland for with environmental and social considerations in mind
monetary factors.

426
JOURNAL OF ALGEBRAIC STATISTICS
Volume 13, No. 1, 2022, p. 426-430
https://publishoa.com
ISSN: 1309-3452

MOTIVATION- NoSQL, Hive, and Mahout-based system that uses

unstructured data and HDFS storage. Wheat and other
Agriculture is critical to India's economy. In recent crops yield numbers were recently released.
years, as a result of industrialisation and the
widespread use of pesticides, the soil's strength has Crops were not given any consideration. A location
deteriorated. Numerous agricultural practises are detection module, data collecting, and analysis are all
insufficient to boost output. The typical challenge part of RSF, according to the research [3]. Database for
faced by Indian farmers is a lack of knowledge on the agricultural cultivation, database for physiographic
appropriate crop depending on their location. Due to data, and module for analysis and storage. As part of
their soil needs, this has an effect on production. Indian the application's location-based mapping feature, a
farmers confront several obstacles while deciding user's present location is compared to a list of similar
which agricultural technology to use and which crop to places and the crops that grow there. A similarity
plant be chosen according to the climate [10][11]. The matrix is used to provide recommendations for the
prevalent issue among Indian farmers is that they do user. The location detecting module uses Google Maps
not choose the correct crops in order to maximise their API services to get the user's current location so that it
yields. Maximise yields depending on topographic and can find similar locations. The system, on the other
economic factors. Maximizing crop yields in hand, does not ask for human feedback to improve the
agriculture [12]. The purpose of manufacturing is to process.
keep costs down.
Majority Voting Technique (MVT) is an ensemble
LITERATURE SURVEY- strategy suggested by S.Pudumalar and co-authors in
paper [4]. Many models are used to increase forecast
These systems take into consideration a wide range of accuracy in this method. Using ensemble methods such
characteristics, including meteorological conditions, as Nave Bayes and Random Trees, KNN, and CHAID,
soil type, terrain, temperature and rainfall, crop market it is possible to assure that no matter how bad one
price, and crop length when making recommendations approach performs, the others will still be accurate. The
for agricultural crops. According to our findings, the majority vote procedure guarantees that the final
following papers were used as sources for our prediction is correct since models are more likely to
investigation and analysis. give accurate predictions.
Users may make better decisions on what crops to plant Prediction relies heavily on the use of if-then logic.
according to a strategy presented by Professor Rakesh The 88 percent model ensemble accuracy is achieved
Shirsath and his co-authors in their research [1]. Each with the use of Several agricultural algorithms and
user's personal information is stored in a subscription- their accuracy are examined in Yogesh's [5] review
based system. Registration has been completed for a study. Gandge and Sandhya. For predicting rice
farmer. The system includes a module that gathers data production, Multiple Linear Regression was shown to
on previous crops grown and recommends a suitable be 90-95% accurate. The ID3 algorithm was used to
crop for planting based on this knowledge. With the analyse and provide recommendations for the soybean
help of artificial neutrals networks, we complete the crop decision-making tree. The third algorithm was
whole process. As a final step, the developer has SVM. A high degree of accuracy was achieved while
included a method for receiving input from the farmer using minimum computational resources for all crops.
so that any issues with the system's functioning may be In order to achieve 95% accuracy, we used a neural
corrected as soon as possible. network trained on corn data. Additionally, Bayes,
There is a lot of data in knowledge databases, naïve, utilised KNN, C4.5, K-means, J48, and LAD
according to Ji-chun Zhao and JianxinGuo [2]. Tree. The conclusion was that further work is needed
Modules such as knowledge engineer, domain expert, to improve the accuracy of the algorithms.
HMI, inference engine, and knowledge base are all Data from Kaggle.com was used in the paper "Use of
instances of users. An effective knowledge foundation Data Mining in Crop Yield Prediction" [6]. Analysis of
for problem resolution is built via the decision system's the data was carried out using WEKA, which is owned
data collection mechanism. The article uses a number by the author. To measure precision, we used
of Hadoop modules to extract characteristics. This is a sensitivity, specificity, accuracy and root-mean-square

427
JOURNAL OF ALGEBRAIC STATISTICS
Volume 13, No. 1, 2022, p. 426-430
https://publishoa.com
ISSN: 1309-3452

error. For each classifier, a confusion matrix was Precise Agriculture Crop Recommendation System [4]
utilised to assess its performance. All of the relevant was developed by S. Pudumalar and E. Ramanujam in
occurrences. It was shown to be possible to acquire 2016. Random tree, CHAID, KNN, Naive Bayes, and
higher accuracy via pruning. WEKA were utilised in this study. Other aspects of this
study include pre-processing the data, dealing with
Seven machine learning strategies were advocated by missing and out-of-range values; feature extraction; an
Rakesh Kumar, J.P. Singh, and Prabhat Kumar in a ensemble model to improve accuracy; and rule
study [7]. ANN, SVM, KNN, Decision Tree, Random development.
Forest, GBDT, and Regularized Gradient Forest are
among the crop selection techniques used. All seeded "A Study on Various Data Mining Techniques for Crop
crops and their development phases at a certain time of Yield Prediction," by Yogesh Gandge, Sandhya, 2017
year are intended to be recovered using this approach. [5]. A decision tree using ID3, SVM, neural networks,
As a result, the best yielding crops are selected for and K-means and KNN, as well as multiple linear
planting. Moreover, the approach advises planting a regression and attribute selection, are some of the
succession of crops in order to maximise output [17]. methods he outlines in this study. In addition,
according to the author's findings, selecting an
COMPARATIVE STUDY: agricultural area, selecting a crop that has already been
On the topic of "Agriculture Decision Support System planted, receiving input from the user, preprocessing,
Using Data Mining," Prof. Rakesh Shirsath worked in selecting attribute values, and running a classification
2017 [1]. A subscription-based system, an Android algorithm on the data are all encouraged in this study.
app, and personalised content were all included in this She is Shruti Mishra.
paper's proposal. The author of this article also focused "Use of Data Mining in Crop Yield Prediction" by
on previously planted crops that the system was aware Priyanka Paygude was published in 2018 [6]. J48,
of in an Android app with a login module. Crop LAD tree, LWL, and IBK algorithms were proposed in
maintenance and user feedback mechanisms. this study. It has been discovered that errors may be
Researchers Ji-chun Zhao and Jian-xinGuo [2] avoided by trimming trees in WEKA's LAD tree,
collaborated on the paper titled "Big Data Analysis whereas IBK was seen to have a greater level of
Technology Application in Agricultural Intelligence accuracy.
Decision System". In this study, he presented an SYSTEM ARCHITECTURE-
Inference engine, Domain expertise, Knowledge
engineering, Knowledge acquisition module, and a Figure 1. System Architecture.
knowledge base for recommendation system usage. A
vast database of crops, which was processed using
Hadoop, was also examined by the author in this work.
Expertise, as well as experiences from the past Future
Scope: Using Hadoop and Artificial Neural Networks
for Feature Selection in HDFS. RSF:

A Recommendation System for Farmers" was the work

of Miftahul Jannat Mokarrama in 2017 [3]. In this
study, he presented a location detection, data analysis,
and storage, similar location detection, and
recommendation generating module. Aside from these
considerations, this research also addressed a number Figure 1 illustrates the flow chart of the system. It
of other aspects of crops, such as the physiographic and describes the whole process starting with the
temperature conditions in which crops grow and the registration and various services provided by the
pace at which they produce, as well as the crop mobile application [16][17].
database.

428
JOURNAL OF ALGEBRAIC STATISTICS
Volume 13, No. 1, 2022, p. 426-430
https://publishoa.com
ISSN: 1309-3452

PROPOSED MODEL- amount of cultivable land available were not taken into
account. These criteria also have a significant impact
Thus, looking upon all these issues that the current on the farmer's profitability, which is the primary
Agro industry is facing, the proposed system helps in reason for a farmer would make crop recommendations
determining the best crop yield and suitable fertilizers using a crop recommendation system.
to be used. Also, it helps to determine the diseases that
plants are affected with. The proposed system is REFERENCES-
divided into 3 parts:
[1] 2017 International Conference on I2C2,
“Agriculture decision support system using data
1. Crop Recommendation System
mining”, Prof. Rakesh Shirsath; Neha Khadke;
a. Best suitable crop that can be grown Divya More.
[2] 2018 the 3rd IEEE International Conference on
b. Top 5 crops as per suitable conditions Cloud Computing and Big Data Analysis, “Big
Data Analysis Technology Application in
2. Fertilizer Recommendation System Agricultural Intelligence Decision System”, Ji-chun
Zhao; Jian-xinGuo.
a. As per soil content and crop give soil quality [3] 2017 IEEE Region 10 Humanitarian Technology
Conference, “RSF: A Recommendation System for
b. Based on soil quality give suggestions and Farmers”, Miftahul Jannat Mokarrama; Mohammad
recommendations to improve it Shamsul Arefin.
[4] 2016 IEEE Eighth International Conference on
3. Plant Disease Detection Advanced Computing (ICoAC), “Crop
Recommendation System for Precision
a. Based on image of the plant classify as Diseased or Agriculture”, S. Pudumalar*, E. Ramanujam*, R.
Healthy and give crop name Harine Rajashreeń, C. Kavyań, T. Kiruthikań, J.
Nishań.
b. As per output of disease, give: [5] 2017 International Conference on Electrical,
Electronics, Communication, Computer and
i. Cause of the disease
Optimization Techniques (ICEECCOT), “A Study
on Various Data Mining Techniques for Crop Yield
ii. Methods to prevent
Prediction”, Yogesh Gandge, Sandhya.
CONCLUSION- [6] Proceedings of the Second International Conference
on Inventive Systems and Control (ICISC 2018),
We concluded from the analysis in this article that “Use of Data Mining in Crop Yield Prediction”,
more research is necessary in the agricultural industry Shruti Mishra, Priyanka Paygude, Snehal
to improve accuracy. Using ensemble approaches helps Chaudhary, Sonali Idate.
assure the system's correctness. Additionally, if we [7] 2015 International Conference on Smart
want to evaluate just one algorithm for the Technologies and Management for Computing,
recommendation system, we may utilise SVM since its Communication, Controls, Energy and Materials
computing needs are minimal. Future study in this (ICSTM), Vel Tech Rangarajan Dr. Sagunthala
topic might include geospatial analysis that R&D Institute of Science and Technology,
incorporates and utilises all available seasonal, soil, Chennai, T.N., India. 6 - 8 May 2015. pp.138-145,
weather, and temperature data. Topography, “Crop Selection Method to Maximize Crop Yield
agricultural production, and farmer economic Rate using Machine Learning Technique”, Rakesh
conditions into a single model in order to produce a Kumar , M.P. Singh , Prabhat Kumar and J.P. Singh
stable and unified system. System accessible to the [8] IEEE Sponsored 9th International Conference on
user. The research shown, in particular, that for Intelligent Systems and Control (ISCO) 2015,
previously suggested recommendation systems, “XCYPF: A Flexible and Extensible Framework for
elements such as the user's investment in agriculture, Agricultural Crop Yield Prediction”, Aakunuri
the number of workers required for upkeep, and the Manjula, Dr. G . Narsimha.

429
JOURNAL OF ALGEBRAIC STATISTICS
Volume 13, No. 1, 2022, p. 426-430
https://publishoa.com
ISSN: 1309-3452

[9] Umamaheswari S, Sreeram S, Kritika N, Prasanth products: State of agricultural systems science.
DJ, “BIoT: Blockchain-based IoT for Agriculture”, Agricultural systems”, 2017 Jul 1;155:269-88.
11th International Conference on Advanced [15] Johnson LK, Bloom JD, Dunning RD, Gunter CC,
Computing (ICoAC), 2019 Dec 18 (pp. 324-327). Boyette MD, Creamer NG, “Farmer harvest
IEEE. decisions and vegetable loss in primary production.
[10] Jain A. “Analysis of growth and instability in the Agricultural Systems”, 2019 Nov 1;176:102672.
area, production, yield, and price of rice in India”, [16] Kumar R, Singh MP, Kumar P, Singh JP, “Crop
Journal of Social Change and Development, Selection Method to maximize crop yield rate using
2018;2:46-66 a machine learning technique”, International
[11] Manjula E, Djodiltachoumy S, “A model for conference on smart technologies and management
prediction of crop yield” International Journal of for computing, communication, controls, energy,
Computational Intelligence and Informatics, 2017 and materials (ICSTM), 2015 May 6 (pp. 138-145).
Mar;6(4):2349-6363. IEEE.
[12] Sagar BM, Cauvery NK., “Agriculture Data [17] Sriram Rakshith. K, Dr. Deepak. G, Rajesh M,
Analytics in Crop Yield Estimation: A Critical Sudharshan K S, Vasanth S, Harish Kumar N, “A
Review”, Indonesian Journal of Electrical Survey on Crop Prediction using Machine Learning
Engineering and Computer Science, 2018 Approach”, In International Journal for Research in
Dec;12(3):1087-93. Applied Science & Engineering Technology
[13] Wolfert S, Ge L, Verdouw C, Bogaardt MJ, “Big (IJRASET), April 2019, pp( 3231- 3234)
data in smart farming– a review. Agricultural [18] Veenadhari S, Misra B, Singh CD, “Machine
Systems”, 2017 May 1;153:69-80. learning approach for forecasting crop yield based
[14] Jones JW, Antle JM, Basso B, Boote KJ, Conant on climatic parameters”, In 2014 International
RT, Foster I, Godfray HC, Herrero M, Howitt RE, Conference on Computer Communication and
Janssen S, Keating BA, “Toward a new generation Informatics, 2014 Jan 3 (pp. 1-5). IEEE.
of agricultural system data, models, and knowledge

430