Professional Documents
Culture Documents
Varaprasad
Varaprasad
Introduction
Domain Knowledge
Problems in existing system
Proposed System
Requirements Specification
Analysis
System Design
Implementation
Testing
INTRODUCTION
The main aim of the project is to predict the soil fertility for better
vegetation cover.
The dataset contains the various elements found in the soil, for instance,
organic matter, various nitrogen compounds, potassium, sodium,
sulphates, boron, etc. It also contains various soil properties like pH.
The target of the data set is to predict the vegetation cover which is the
percent vegetative cover of the 102m area surrounding the sample site.
The higher the vegetation cover higher is the fertility of the soil for crops.
Vegetation cover is calculated in percentage from 1 to 100, so, it becomes
the regression task. To achieve the results various regression methods are
applied and performance of each model is analysed.
REQUIREMENT SPECIFICATION
Software Requirements :
Operating System : Windows
Language : Python
Dataset : MS Excel (CSV – “Comma-Separated Value”)
Software IDE : Google Collaboratory
Web Browser
Documentation : Microsoft Office
Hardware Requirements :
Processor : i5 (Recommended)
RAM : 4GB (Recommended)
Hard Disk
REQUIREMENT SEPECIFICATION
Functional Requirements :
Collection of soil attribute data set.
Data Pre-processing
Data Visualization
Working on various machine learning algorithms to find the more
accurate one.
The system can predict soil fertility.
The main aim of the analysis phase in software development life cycle is
• To identify the implementation details.
• To understand the architecture of the system.
• To identify what the system is supposed to do.
Random forest regression is a machine learning algorithm that belongs to the family of
decision tree-based algorithms.
It can handle a large numbers of features and can scale well to large datasets.
It can provide useful insights into the importance of different features of making
predictions
RANDOM FOREST REGRESSION (Contd.,)
24.Mg(meq/100g) 25.Na(meq/100g)
PREPROCESSED DATASET
Soil attribute dataset contains various nutrients values that are required to maintain
the fertility of the soil.
It also contains the vegetation cover of the soil which defines the how much fertile
the soil is.
After removing unnecessary data, Preprocessed Dataset contains various attributes
like :
Test Case 2 :
Input :
NO3=1.75 NH4=0.5 P=17 SO4=4 K=6 B=0.25 Organic Matter=0.10
pH=8.8 Zn=0.3 Cu=0.1 Fe=1 Ca=4.7 Mg=0.5 Na=0.1
Output: 14
CONCLUSION:
The system uses supervised Machine learning algorithms like Linear Regression, Decision
tree algorithm, Support Vector Machine, Random Forest Classifier and gives best result
based on error analysis. The results of these algorithms will be compared and the best
among them i.e., Random Forest Classifier which gives the best and accurate output is
chosen.
FUTURE SCOPE:
In this project, we give suggestions to user when the user gets less percentage of
fertilily based on the nutrient values they provide. So In the future, we aim to
give suggestions to the user based on the crop that user needs to cultivate by
comparing the nutrient values of various crops dataset and also we add some
input fields to get information related to the crop that user want to cultivate.