Crop Yield Prediction Using Machine Learning
Crop Yield Prediction Using Machine Learning
Volume 9, Issue 2, February – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Crop Yield Prediction Using Machine Learning
1
Meka.Viswas
2
Kedarisetti.Sri.Naga.Veera.Venkata.Swami.Subha.Surya
3
Kathula.Ramu
4
Medapati. Rama.Phaneendra Reddy-,
Swarnandhra College of Engineering &Technology.
Abstract:- Agriculture is an important part of the Indian many seeds. This shows that predicting crop yield is a result of
economy and more than half of the country's population the development process and not an explicit interaction.
earns their living from agriculture. Agriculture is an Potential models can now easily calculate actual outcomes;
important part of the growth of human civilization but better production forecasts are still needed. Regression
through the raising of domesticated animals that produce techniques were used to make future predictions, while
food that enables people to survive. Machine learning is descriptive models were used to gain insight from the
used to predict crop yield based on parameters such as collected data and explain the results. Mechanical engineering
rainfall, yield and weather. In addition to being an research presents many challenges when trying to develop
important decision-making tool for crop yield prediction, high-performance models. Choosing the right algorithm for
machine learning also supports crop production and crop the job is crucial; The algorithm and underlying platform need
production-related decision-making. commonly used to be able to handle a lot of data.
algorithms It is a neural network device. Weather, climate
and other environmental factors pose a long-term threat to In the Indian economy, the main occupation of people
agriculture. Machine learning (ML) is important because living in rural areas is agriculture. According to 2018
it provides decision support tools for crop forecasting statistics, its population makes up more than half of the
(CYP) that can help make decisions such as which crops to country, but its GDP accounts for only 17% of the country.
plant and how during the growing season. The main Today's farmers grow products based on centuries of
limitation of neural networks is to reduce the relative error experience. Since the time of traditional methods, there have
and efficiency of crop yield prediction. The main objective been examples of such over- or under-production where the
of crop forecasting is to improve crop production and actual rules for growing a crop are not observed. This is a
various models are used to achieve this goal. This research major source of stress for Indian farmers. Farmer suicides are
helps make agriculture more efficient by demonstrating increasing due to low productivity, inbreeding and high
machine learning's ability to predict crop yields with high production costs. All over the world, soil is being lost, water is
levels of productivity. Design can be a decision support polluted, and farmers' incomes are slowly decreasing. This
tool for farmers, enabling them to make informed problem can be solved by trying new crops, but farmers have
decisions on crop management, resource allocation and difficulty predicting the results of a single crop. If they can
risk mitigation, ultimately increasing agricultural predict which crop will yield the most, they will make more
sustainability and food security. Using the results of this money and support farmers. Forecasting is one of the latest
study, farmers will be able to make informed decisions by changes in agriculture and has a huge impact on the industry.
determining the yield of their crops before planting on It uses satellite imagery, comprehensive data on soil
their farms. parameters, weather conditions and crop history to achieve
optimum results. Therefore, we chose to use the most effective
Keywords:- Crop_Yield_Prediction; Logistic_Regression; regression models, compare their accuracy, and see which
Naive Bayes; Random Forest; Dataset. model best predicted yield when some weather and soil
parameters were not entered.
I. INTRODUCTION
India grows crops throughout the year. Crops are
Machine learning (ML) technology is used in everything classified according to growing season. There are two growing
from predicting mobile phone usage to analyzing consumer seasons in India: Kharif season and Rabi season. Kharif and
behavior in supermarkets. Machine learning has been used in Rabi growing season is from July to October and October to
agriculture for a long time. Predicting crop yield is one of the March. Thereafter crop and weather data were combined into
most difficult problems in horticulture, and to date there are a single file. Crop history information such as crop region,
many models to be applied and evaluated. This question name, season, and other climatic factors affecting the crop are
requires a lot of data because the harvest is affected by many used as independent variables. The yield will be extraordinary.
factors such as soil, weather, environment, use of compost and
IJISRT24FEB1278 www.ijisrt.com 1245
Volume 9, Issue 2, February – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
II. LITERATURE SURVEY as India. The use of age records in agriculture can change
the decision-making environment and farmers can earn
Madhuri Shripathi Rao[1] said that agriculture is the main higher profits. Agriculture, which supports about a fraction
source of income for many developing countries. Today's of India's population, accounts for only 14% of India's
agriculture is a path to farming success and continuous GDP. A plausible explanation behind this is that farmers
improvement of agricultural technology. It is important for do not have sufficient judgment regarding yield estimates.
farmers to respond to changes in the world, traders, There is no basis for telling farmers about the plants they
consumers, etc. It becomes difficult to meet their will grow. Machine learning techniques, rainfall,
expectations. Some of the problems faced by farmers are: temperature, humidity, etc. It aims to predict the best crop
(i) Tackling climate change due to soil erosion and in a particular region by analyzing various atmospheric
industrial emissions (ii) Inadequate soil nutrients, shortage factors such as.
of essential nutrients such as potassium, nitrogen and Narasimhamurthy, V[5] said that Rice is India's most
phosphorus in vegetables slow down the harvest. important crop after wheat. India ranks second in the world
A.M., Abouelghar [2] said that Pre-harvest forecasting of after China and is generally considered a major producer
crops can prevent bad situations and help decision makers of rice, accounting for 20% of the world's total production.
use reliable and accurate strategies on food safety. Remote The cultivated area in 20 states of India is approximately
sensing has many awards in the field of crop monitoring 40 million hectares. India continues to be the world's
and yield prediction related to soil, climate and biophysical largest exporter with exports exceeding 100 million tons.
and biochemical changes. Different technologies are Sustainability and efficiency of growing crops under
available for crop monitoring and yield prediction, favorable conditions. Developing better methods for
including multispectral and hyperspectral data, radar and predicting crop productivity in different climates can help
lidar imagery. farmers and other stakeholders make better decisions in
Ms. Pratistha Mathur [3] said that Agriculture is an agronomy and crop selection.
important part of the economy in India. Population growth
is a problem for food security. Population growth leads to III. METHODOLOGY
increased demand, which requires farmers to use the same
farmland to produce more crops to increase capacity. This work aims to develop a regression algorithm using
Technology can help farmers increase crop yields by Google Collaboration Tools and Jupyter Notebook. The
predicting crop yields. The main objective of this article is modeling steps are shown in the figure. The first step is to get
to estimate crop yield using area, yield, yield and irrigated the dataset from the Kaggle website; then before
area. Four machine learning methods decision tree, linear documentation, including maintenance and construction
regression, lasso regression and ridge regression were used documentation; then perform data analysis (EDA); For some
to predict the yield. Cross-validation method was used for models, machine learning and random forest algorithms
validity, and absolute error, mean square error and root should be used. information. The table below shows some
mean square error were used for validity. Decision trees information about different laptops and their prices according
outperform other machine learning methods. to their features. Data from Kaggle.com
Sangeeta, S.G[4] said that Agriculture is an important
source of energy, especially in developing countries such
Fig 1: Sample dataset for crop yeild
IJISRT24FEB1278 www.ijisrt.com 1246
Volume 9, Issue 2, February – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
This data can cleaned and explorated using machine learning techniques and that mostly sutable forRandom forest algorithms .
The data flow can be represented as follows.
Fig 2 : data flow of model
Random Forest algorithm can be applied on the the dataset for the classification(To get the optimal price from the dataset). By
using the random forest algorithm works 85% accurately.
Fig 3: Random Forest
Random forests work in two stages; first, it creates a random forest by combining N decision trees, and second, it makes
predictions for each tree created in the first stage.
We can explain the working process in the following steps and ways:
Step 1: Select the content K data from the training process.
Step 2: Create a decision tree with the selected data points (subset).
Step 3: Select the number N for the decision tree you want to create.
Step 4: Repeat steps 1 and 2.
Step 5: For new data, find the prediction for each decision tree and put the new data points in the class that won the most votes.
IJISRT24FEB1278 www.ijisrt.com 1247
Volume 9, Issue 2, February – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
IV. APPLYING RANDOM FOREST
For this dataset we got the accuracy upto 85% by using random forest algorithm.
Fig 4: working code of model.
V. CONCULUSON [4]. Sangeeta, S.G.: Design and use of predictive models
used to produce crops. internationalization. J. education.
The data we use in this analysis is provided by Kaggle machine. resources. 8(01) (2020)
and is based on facility data. In this analysis, we perform data [5]. Narasimhamurthy, V.: Rice yield prediction using
cleaning, data analysis and data visualization. We concluded random forest algorithm SML. internationalization. J.
that product prediction can be made with 85% accuracy by Res. application form. education. engineer. machine. 5,
using random forests and machine learning techniques for 1220–1225 (2017).
classification.
REFERENCES
[1]. Crop Forecasting Using Machine Learning Madhuri
Shripathi Rao1, Arushi Singh1, N.V. Subba Reddy1 and
Dinesh U Acharya1 1Department of Computer Science
and Engineering, Manipal Institute of Technology,
Manipal, 576104, Udu, Karnataka, India Pi Region,
2022.
[2]. Ali, A.M., Abouelghar, M.A., Belal, A.-A. , Saleh, N.,
Younes, M., Selim, A., et al. (2022). Crop yield
estimation using multiple remote sensing (see article).
Sweetcorn. J. Remote Sensing Space Science. 25 Ib.,
711-716. doi:10.1016/j.ejrs.2022.04.006
[3]. Kavita, Ms. Pratistha Mathur, Crop yield prediction
using machine learning in India 2020 IEEE 5th
International Conference on Computational
Communications and Automation (ICCCA) (2020),
