A Seminar Report

on
“Fertilizer recommendation System”

Submitted to the
Savitribai Phule Pune University
In partial fulfillment for the award of the Degree of
Bachelor of Engineering
in
Information Technology
By

Sujal Jagtap[TIA19]

Under the guidance of

Prof. Priti Yampe

Department Of Information Technology

International Institute of Information Technology, Pune

Hinjewadi, Pune , Maharashtra, India
2024-2025
 CERTIFICATE

This is to certify that the Seminar report entitled “Fertilizer recommendation System” being
submitted by Sujal Jagtap is a record of bonafide work carried out by him/her under the supervision
and guidance of Prof. Priti Yampe in partial fulfillment of the requirement for TE (Information
Technology Engineering) – 2019 course of Savitribai Phule Pune University, Pune in the academic
year 2024-2025

Date:

Place:

Prof. Priti Yampe Dr. Jyoti Surve

Guide Head of the Department

This Seminar report has been examined by us as per the Savirtibai Phule Pune University, Pune,
requirements at International Institute of Information Technology, Pune

(Name & Signature) (Name & Signature)

Internal Examiner External Examiner

ACKNOWLEDGEMENT
With immense pleasure we are presenting this project report on “ Fertilizer recommendation System”
as a part of Seminar Presentation TE Information technology at INTERNATIONAL INSTITUTE OF
INFORMATION TECHNOLOGY ,HINJEWADI ,PUNE.

It gives us the privilege to complete this report under the valuable mentorship of Prof. Priti Yampe.
Her guidance, cooperation, and encouragement have made headway in the project.
We are also extremely grateful to Dr. Jyoti Surve (H.O.D of Information Technology) and Dr. Vaishali
Patil, Principal, Information Technology of Information Technology for providing all facilities and help
for the smooth progress of the report work. We would also like to thank all the Staff Members of the
Information Technology Department, Management, friends, and our family members, who have
directly or indirectly guided and helped us with the preparation of this report and gave us unending
support right from the stage the idea was conceived.

Students Name & Signature

Sujal Jagtap
 Abstract

I am deeply grateful to all those who have contributed in various ways to the successful completion of
this project titled "Creation of a Data-Driven Procedure for the Best Fertilizer Recommendation".
This work has been the result of considerable collaboration and guidance, and I take this opportunity
to express my heartfelt thanks.

First and foremost, I wish to extend my profound gratitude to my mentor, Prof Priti Yampe, whose
expertise, thoughtful advice, and unwavering support have been invaluable throughout the journey of
this research. Her insightful feedback and meticulous guidance have shaped this project into what it is
today. I am forever indebted to her for the countless hours she has devoted to helping me refine my
work.

Additionally, I would like to offer my sincere appreciation to Dr . Jyoti Surve for her invaluable help
and suggestions. Her constructive feedback and consistent encouragement played a significant role in
improving the quality of this research, and her advice on specific aspects of the project has been
indispensable.

I am also grateful to the academic staff of the Information Technology department at my institution
for providing the essential academic framework and moral support throughout the process. Their
guidance, whether formal or informal, has contributed significantly to the project’s success.

My appreciation further extends to the technical team, whose assistance with data analysis software
tools and methodologies was crucial in the execution of the research. Their timely help and support in
navigating technical challenges enabled smoother progress and ensured that the data-driven approach
was effectively realized.
 Table of Content

1. Introduction 6-7
1.1 Purpose
1.2 Scope
1.3 Intended Audience
2. Literature Survey of Plant Disease Detection 8-9
2.1 Introduction
2.2 Different Approaches
3. Machine Learning Approach 10
3.1 Introduction
3.2 Product functions
3.3 User Characteristics
3.4 Principle Actors
3.5 Assumptions and Dependencies
3.6 Assumptions and Dependencies
4. Application Of Fertilizer Recommendation Model: 11
5. Algorithm Used : Decisition Tree 12-13
6. Steps of Implementation 14-15
7. Result Discussion 16-18
8. Advantages And Disadvantages 19
9. Conclusion 20
10. References 21

.
 CHAPTER 1

INTRODUCTION TO FERTILIZER RECOMMENDATION

SYSTEM
1.1 Introduction to fertilizer recommendation system Using ML

The Fertilizer Recommendation System focuses on addressing a major issue in modern agriculture:
the overuse and improper application of fertilizers, which leads to soil degradation and reduced
agricultural productivity. This project aims to develop a data-driven solution using machine learning
to provide personalized fertilizer recommendations to farmers, based on factors like soil health, crop
type, and weather patterns. By optimizing fertilizer use, the project seeks to increase crop yields,
improve farmers’ incomes, and promote sustainable agricultural practices.

1.2 Scope

This project aims to develop a data-driven solution that optimizes fertilizer recommendations based
on key factors such as soil health, crop type, and environmental conditions. By analyzing large
datasets from agricultural research and real-world farming practices, the system provides tailored
fertilizer recommendations that promote sustainable farming, enhance crop yield, and improve soil
quality. The solution integrates machine learning algorithms and data analytics to deliver precise and
context-specific recommendations, ensuring the responsible use of fertilizers while addressing the
problem of excessive or improper fertilizer application.

1.3 Intended Audience

The intended audience for this document comprises developers, data scientists, agricultural experts,
and stakeholders involved in the development, deployment, and utilization of the Fertilizer
Recommendation System.

Aim : This project is to develop an efficient and accurate system that provides personalized fertilizer
recommendations to farmers using data-driven techniques and machine learning algorithms. This
system will assist farmers in optimizing fertilizer application based on critical factors such as soil
health, crop type, and weather conditions, ultimately improving crop yields, reducing input costs, and
promoting sustainable agricultural practices. By leveraging advanced technologies, the project seeks
to minimize the environmental impact of improper fertilizer use and enhance overall agricultural
productivity.

Project objectives: The Fertilizer Recommendation System project aims to create an efficient
platform that utilizes machine learning algorithms to provide accurate and personalized fertilizer
recommendations to farmers. The system will analyze crucial factors such as soil health, crop type,
and weather patterns to optimize fertilizer use. By developing a user-friendly interface, the project
will enable farmers to input relevant data and receive real-time recommendations tailored to their
specific needs. Utilizing large agricultural datasets, including soil and crop data, the system will
improve its accuracy over time through continuous learning.
The Fertilizer Recommendation System is designed to optimize fertilizer use by providing
personalized recommendations to farmers based on advanced data analysis and machine learning
techniques. By analyzing key factors such as soil health, crop type, and weather patterns, the system
delivers precise fertilizer recommendations that help farmers improve crop yields and reduce input
costs. This data-driven approach eliminates guesswork and ensures more efficient and
environmentally friendly fertilizer application..
 CHAPTER 2

LITERATURE SURVEY OF FERTILISER RECOMMENDATION

SYSTEM
Crop Yield Prediction and Efficient use of Fertilizers

INDIA is a highly populated country and randomly change in the climatic conditions need to secure
the world food resources. Framers face serious problems in drought conditions. Type of soil plays a
major role in the crop yield. Suggesting the use of fertilizers may help the farmers to make the best
decision for their cropping situation. The number of studies Information and Communication
Technology (ICT) can be applied for prediction of crop yield.

Prediction of Crop Fertilizer Consumption

Amount of fertilizers consumption is calculated on the basis of the area of the deficient region for
nitrogen deficient plant. It was observed that nitrogen deficiency was seen at early stage of plant
growth when compared to later stages.

Use of Various Machine Learning Algorithm

The application of machine learning (ML) has gained significant traction in recent years. Support
Vector Machines (SVM), Decision Trees, Logistic Regression, and Random Forest algorithms have
been employed to classify fertilizers based on features extracted from soil data. It was concluded from
this literature review that MapReduce should be used as an efficient programming model for
computing very large datasets of weather and climate with ease and high performance

Deep Learning Approaches

Algorithms like Decision Trees, Random Forests, and Support Vector Machines have been used to
analyze various factors, including soil health, crop type, and environmental conditions. However,
deep learning models, particularly those incorporating neural networks, offer the ability to process
complex patterns in data with greater accuracy. These models, when trained on large datasets of soil
characteristics and crop growth patterns, can deliver highly personalized recommendations.

Challenges and Future Directions

Despite the advancements in data-driven fertilizer recommendation systems, several challenges

remain. One of the most pressing challenges is environmental variability, which makes it difficult
for models to generalize across different regions and climates. Moreover, the complexity of soil
composition and the need for high-quality, localized datasets continue to pose obstacles to
achieving maximum accuracy.
TABLE 1. LITERATURE SURVEY
YEAR AUTHOR TITLE METHODS KEY
FINDINGS
2020 Snehal S. "Agricultural Crop Yield Artificial Used neural
Dahikar et al. Prediction Using Neural networks for
Artificial Neural Networks predicting
Networks" agricultural
crop yields.
2018 Anna "Machine Learning Machine Reviewed
Chlingaryana et Approaches for Crop Learning various
al. Yield Prediction and machine
Nitrogen Status learning
Estimation in Precision techniques for
Agriculture" crop yield and
nitrogen
estimation.
2018 A.T.M. Shakil "Applying Data Mining Data Mining Predicted crop
Ahamed et al. Techniques to Predict fertilizer using
Annual Yield of Major various data
Crops in Bangladesh" mining
techniques.
2016 K.E. Eswari, L. "Crop Yield Prediction Bayesian Applied
Vinitha in Tamil Nadu Using Network Bayesian
Bayesian Network" networks to
predict crop
yield in Tamil
Nadu.

2014 Shriya Sahu et "Efficient Analysis of Hadoop, Implemented a

al. Crop Yield Prediction Random Forests crop fertilizer
Using Hadoop prediction
Framework Based on framework
Random Forest using Hadoop
Approach" and Random
Forests.
 Chapter 3
Machine Learning Approach

1. Introduction to Fertilizer Recommendation System Using ML

The Fertilizer Recommendation System is a standalone application designed to operate
independently. It interfaces with users through a web-based interface, enabling seamless interaction
for inputting soil data and receiving personalized fertilizer recommendations. The core functionality
of the system is based on machine learning models implemented using TensorFlow, which ensures
precise and efficient fertilizer recommendations based on soil health, crop type, and environmental
factors.

2. Product Functions
Training a model for fertilizer recommendation involves developing a machine learning model
capable of analyzing various factors such as soil properties, crop requirements, and weather patterns
to provide suitable fertilizer suggestions. Below is a step-by-step guide covering the essential
aspects of training such a model:
• Users: Farmers and agricultural experts providing soil and crop data.
• System: The Fertilizer Recommendation System processes data and provides
optimized fertilizer recommendations.

3. User Characteristics
• Farmers: Typically possess limited technical knowledge but are proficient in
agricultural practices and soil management.
• Agricultural Experts: Familiar with technology and interested in utilizing AI to
enhance the accuracy of fertilizer application.
• Developers: Have technical expertise in machine learning, data processing, and
software development.

4. Principal Actors
• Users: Farmers and experts inputting soil data and requesting fertilizer
recommendations.
• System: The Fertilizer Recommendation System responsible for analyzing the input
data and generating recommendations based on the trained model.

5. Assumptions and Dependencies

• Data Availability: Availability of labeled datasets, including soil characteristics, crop
types, and fertilizer requirements for training and testing the model.
• Computational Resources: Access to computational resources for training and
deploying machine learning models, such as CPUs and GPUs.
• Software Dependencies: Dependence on TensorFlow and other libraries for
implementing machine learning algorithms necessary for the system to function
efficiently.
 CHAPTER - 4
Application Of Fertilizer Recommendation Model:
With advancements in machine learning, data analytics, and sensor technologies, fertilizer
recommendation systems are increasingly integrated into precision agriculture and smart farming
systems. These models enable more efficient, sustainable, and profitable farming practices. Below
are some key applications of fertilizer recommendation models:
1. Precision Agriculture:
Optimized Nutrient Management: Fertilizer recommendation models analyze soil data, crop type,
and environmental conditions to provide customized fertilizer suggestions, ensuring that plants
receive the right nutrients in the right amounts at the right time.
Targeted Fertilizer Application: By identifying the specific nutrient requirements of different areas
within a field, farmers can apply fertilizers more accurately, reducing waste and costs while
improving soil health.
2. Improved Crop Yield and Quality:
Enhanced Crop Growth: Providing the correct type and amount of fertilizer based on real-time soil
analysis helps ensure optimal plant growth, leading to increased crop yield and improved produce
quality.
Soil Conservation: Fertilizer models promote sustainable farming practices by preventing over-
fertilization, which can degrade soil quality over time. This results in long-term soil fertility and
environmental sustainability.
3. Automated Farm Management Systems:
Smart Equipment Integration: Fertilizer recommendation models can be integrated into automated
machinery such as smart tractors or drones, which can analyze soil composition in real-time and apply
fertilizers only where needed, reducing labor costs and improving efficiency.
Smart Irrigation and Fertigation Systems: When combined with environmental and soil data, these
models can be linked to smart irrigation systems that not only manage water supply but also deliver
fertilizers through irrigation (fertigation), ensuring precise nutrient distribution.
4. Decision Support Systems for Farmers:
Mobile and IoT-based Solutions: Farmers can use mobile apps integrated with fertilizer
recommendation models to input soil and crop data and receive real-time fertilizer recommendations.
These solutions can also use IoT devices like soil sensors to continuously monitor soil conditions and
adjust fertilizer application in response to changes.
Weather and Environmental Data Integration: Fertilizer models, when combined with real-time
weather data, can offer dynamic recommendations based on changing environmental conditions (e.g.,
rainfall or temperature), helping farmers adjust their fertilization practices to optimize crop growth
and prevent nutrient loss.
 CHAPTER 5

Algorithm used: Decision Tree Algorithm

Decision Tree Algorithm
Introduction: A Decision Tree is a supervised machine learning algorithm used for both
classification and regression tasks. It works by recursively splitting the dataset into subsets based on
the most significant feature at each node, creating a tree-like model of decisions. The final prediction
is made by following the path from the root node to a leaf node, which contains the output class or
value.

Key Concepts:
1. Root Node: The top node representing the entire dataset, which is split into two or more
homogeneous sets based on a feature.
2. Decision Nodes: Intermediate nodes that represent decisions based on splitting the data further
using certain feature values.
3. Leaf Nodes (Terminal Nodes): The end nodes where a class label (for classification) or a value
(for regression) is assigned.
4. Splitting Criteria:
• Gini Impurity: Used in classification to measure how often a randomly chosen
element would be incorrectly labeled.
• Entropy (Information Gain): Measures the disorder or uncertainty in the dataset.
• Variance Reduction: Used in regression to minimize the variance within each split.
Algorithm Steps of Decision Tree

1. Step 1: Select the best feature to split the data based on splitting criteria like Gini
Impurity, Entropy, or Variance.

2. Step 2: Split the dataset into subsets based on the selected feature and its value(s).

3. Step 3: Recursively repeat the process for each subset to create decision nodes until:

• A stopping criterion is met (e.g., maximum depth, minimum samples per

node).

• No further splits are possible (i.e., all instances have the same label, or the
node cannot be split further).

4. Step 4: Assign a class label (classification) or value (regression) to each leaf node.

5. Step 5: For a new data point, traverse the tree from the root node based on feature
values until reaching a leaf node, which provides the final prediction.

Working Principle:

A Decision Tree works by splitting the dataset into smaller subsets based on the most significant
feature, determined using criteria like Gini Impurity or Information Gain for classification, and
variance reduction for regression. At each decision node, the algorithm selects the best feature to
minimize uncertainty or impurity. This process continues until the stopping criteria are met,
resulting in a tree where the final predictions are made at the leaf nodes.
 Chapter 6
Steps of Implementation of “Fertilizer recommendation System”
Step 1: User Registration and Login
• Navigate to the homepage of the web application.
• Select the "Register" option if you are a new user or the "Login" option if you already have an
account.
• For new users, fill out the registration form by providing essential details such as name, email,
password, and other relevant information.
• After successfully registering, log in using your credentials (email and password).
• Once logged in, you will be redirected to the user dashboard, where further actions can be
taken based on the application's functionality.

Step 2: User Exploration of Application

• After successfully logging in, users are directed to the dashboard.
• From here, users can navigate through the various features of the application, such as
exploring available tools, data visualizations, and other functionalities.
• The sidebar or top navigation menu provides easy access to different sections like Dashboard,
Features, User Profile, and Help.

User Interface

Step 3: User Input of Nutrient Data and Crop Name

• After exploring the application, users are prompted to input essential data for fertilizer
recommendations.
 • Users will be asked to provide the amount of Nitrogen (N), Phosphorus (P), and Potassium
(K) in the soil, along with the name of the crop they are growing.
• The input form will typically have labeled fields for entering the values of nitrogen,
phosphorus, and potassium, as well as a dropdown or text input for selecting or typing the
crop name.

User Input

Step 4: Output Display of Fertilizer Recommendations

• After submitting the input data (Nitrogen, Phosphorus, Potassium values, and crop name), the
system processes the information.
• The final fertilizer recommendation, customized to the user's crop and soil data, is displayed
on the screen.
• The recommendations may include the exact type and amount of fertilizer to apply, along
with additional guidelines for proper usage and timing.
• Users can also view insights or suggestions to optimize fertilizer use and improve crop health
based on the provided data.

Output Display
 Chapter 7
Result Discussion
1) Performance Evaluation:
Accuracy: Accurate prediction of nitrogen deficiency in plants is critical for ensuring that farmers
apply the correct amount of fertilizer. By analyzing that data, the system ensures that nutrient
deficiencies are identified with precision. This level of accuracy helps avoid both under- and over-
fertilization, leading to healthier crops and better yields.
Efficiency: The efficiency of the system lies in its ability to quickly process data entered by users and
provide real-time predictions. By optimizing algorithms for data processing, the system is able to
detect nitrogen deficiencies in a timely manner, allowing farmers to take immediate corrective
actions. This efficiency is crucial for minimizing crop damage and optimizing fertilizer use.
Scalability: As agricultural systems grow and the demand for precision farming tools increases,
scalability becomes a key factor. The system’s ability to handle large volumes of data and provide
accurate predictions across different stages of plant growth demonstrates its capacity to scale.
Ensuring that the system can accommodate more users and larger datasets without compromising
performance is vital for widespread adoption.
2) Usability and User Experience:
User Interface: A user-friendly interface plays a significant role in enhancing the system’s
accessibility and usability. The interface should allow users, such as farmers and agricultural experts,
to easily provide input and receive accurate fertilizer predictions without needing technical expertise.
A clean, intuitive design ensures that the system can be used by individuals with varying levels of
familiarity with technology.
Accessibility: Ensuring that the system works across a variety of devices, including smartphones and
tablets, makes it accessible to farmers in different locations. The ability to use the system on standard
hardware configurations means that users do not need specialized tools or equipment to benefit from
the system’s capabilities. Accessibility is key to ensuring that the system can be widely adopted in
various agricultural settings.
3) Implications for Agriculture:
Fertilizer Optimization: One of the most significant benefits of the system is its ability to optimize
fertilizer usage. By accurately predicting when and how much fertilizer is needed based on the nutrient
status of the plant, the system helps farmers avoid wasting resources. This reduces the likelihood of
over-fertilization, which can lead to nutrient runoff and environmental damage, while also ensuring
that plants receive the right amount of nutrients to thrive.
Precision Agriculture: The integration of fertilizer prediction into agricultural practices supports the
concept of precision agriculture, where decisions are based on data-driven insights. By providing
real-time information on the nutrient needs of crops, the system allows farmers to apply fertilizers in a
more targeted and efficient manner. This leads to better resource management, higher crop yields, and
reduced environmental impact.
Sustainable Farming Practices: The system’s ability to minimize fertilizer waste and reduce reliance
on chemical inputs promotes sustainable agriculture. By enabling early detection of nutrient
deficiencies and providing tailored fertilizer recommendations, the system helps farmers adopt more
environmentally-friendly practices. This contributes to the long-term sustainability of agricultural
systems by reducing the negative impact of excessive fertilizer use on the environment.
Imports:

Feature extraction

Training Model

Dataset usage
Accuracy Compaarison
 Chapter 8

Advantages of the Fertilizer recommendation System:

1 Optimized Fertilizer Use:
The system provides precise recommendations for the type and amount of fertilizer needed based on
real-time soil and crop data, ensuring that nutrients are applied in the correct proportion, reducing
waste.

2 Increased Crop Yield:

By ensuring that crops receive the optimal nutrients at the right time, the system promotes healthy
growth, leading to better crop yields and higher-quality produce.

3 Sustainable Farming Practices:

The system minimizes the overuse of fertilizers, which can lead to soil degradation and
environmental harm. It encourages sustainable use of fertilizers, promoting long-term soil health.

4 Cost-effective:
Reducing over-fertilization cuts down on fertilizer costs, which is beneficial for farmers, especially
smallholders who may have limited resources.

5 Data-Driven Decisions:
Farmers receive fertilizer recommendations based on soil composition, crop type, and
environmental conditions, enabling data-driven decision-making for better farm management.

Disadvantages of the Fertilizer recommendation System:

1. Dependence on Data Accuracy:

The system’s effectiveness relies heavily on accurate soil and environmental data. If the data
entered by the user or collected from sensors is incorrect or incomplete, it can lead to inaccurate
recommendations.

2. Limited to Known Crops and Conditions:

The system may not provide accurate recommendations for crops or soil types not included in
the training data, limiting its effectiveness in regions with unique agricultural conditions.

3. Initial Development and Setup Costs: Factors:

Developing and deploying an effective fertilizer recommendation system requires significant
initial investment in machine learning models, datasets, and infrastructure, making it costly to
build and maintain.
 CONCLUSION
The Fertilizer Recommendation System represents a significant advancement in agricultural
technology, addressing the critical challenges faced by farmers in optimizing fertilizer use. As
agriculture continues to evolve amidst growing concerns about sustainability and environmental
impact, this system provides a vital solution by harnessing the power of data analytics and machine
learning.

By systematically analyzing key factors such as soil health, crop type, and weather patterns,
AgriNexus empowers farmers with precise, personalized fertilizer recommendations tailored to their
unique conditions. This not only enhances crop yields and minimizes input costs but also promotes
responsible fertilizer use, mitigating the risks of overapplication and its detrimental effects on soil
health and the environment.

The system's reliance on large datasets ensures that it remains relevant and accurate, continuously
learning from real-world applications and adapting to changing agricultural landscapes. This capacity
for ongoing improvement positions AgriNexus as a valuable tool for farmers, facilitating timely and
informed decision-making that can lead to increased productivity and profitability.

Furthermore, the project's commitment to sustainable agricultural practices aligns with global efforts
to promote environmentally friendly farming. By reducing reliance on manual inspections and
guesswork in fertilizer application, the AgriNexus system contributes to more efficient agricultural
operations and supports the broader goal of sustainable food production.

In conclusion, the AgriNexus – Fertilizer Recommendation System not only addresses the
immediate needs of farmers for effective fertilizer management but also contributes to the long-term
sustainability of agricultural practices. By leveraging advanced technology to optimize resource use,
it stands as a transformative solution that can drive positive change in the agricultural sector,
ultimately benefiting farmers, consumers, and the environment alike.
 REFERENCES

1] Komal Abhang, "Soil Analysis and Crop Fertility Prediction," vol. V, no. 03, 2018.
2] Shriya Sahu et al," An Efficient Analysis of Crop Yield Prediction Using Hadoop framework based
on random Forest approach, International Conference on Computing, Communication and
Automation , 2017.
3] Jig Han Jeong et al , “Random Forests for Global and Regional Crop Yield Predictions ”,PLOS-
ONE DOI:10.1371/journal.pone.0156571 June 3, 2016
4] D Ramesh, B Vishnu Vardhan, "Analysis of crop yield prediction using data mining techniques",
International Journal of research in engineering and technology,eISSN:2319-1163.
5]Comparison of Various Machine Learning Techniques for fertilizer recommendation
6] Machine Learning by Tom M. Mitchell, ISBN: 0070428077
7] K.E. Eswari. L.Vinitha. " Crop Yield Prediction in Tamil Nadu Using Baysian Network ",
International Journal of Intellectual Advancements
8] S. C. P. C. S. G. Komal Abhang, "Soil Analysis and Crop Fertility Prediction," vol. V, no. 03,
2018. [4] S.Y.Bansode , and S.M. Pujar, “Cloud Computing and Libraries.” DESIDOC Journal of
Library & Information Technology, Vol. 32, No. 6, November 2012, pp. 506-512
9] Crop Dataset : (https://www.kaggle.com/datasets/atharvaingle/crop-recommendation-dataset)
10]Kaggle Dataset Link: (https://www.kaggle.com/discussions/accomplishments/473870)
11] Refered youtube videos: ( https://youtu.be/x_SLT4VqpdE?si=A_HWv7UiEUzHGVWF)